Digitalization is advancing, also in NIR spectroscopy, which enables trainable miniature smart sensors e.g. for analyses in the food&feed, chemical and pharmaceutical sectors.
The calibration is the core of a NIR spectroscopy sensor, it enables the numerous applications and should therefore not be the weakest link in the measurement chain.
The development of calibrations that turn NIR spectrometers into smart sensors is done manually by experts (NIR specialist, chemometrician, data scientist) with so-called chemometrics software.
This is very time-consuming (time to market) and the result is person-dependent and thus suboptimal, because each expert has his own preferred way of proceeding.
In addition, the calibrations have to be maintained, as new data has been collected in the meantime, which can be used to extend and improve the calibrations.
This is where our automated service comes in, combining the knowledge and good practices of NIR spectroscopy and chemometrics collected in one software and using machine learning to generate optimal calibrations.
Based on this, we have developed a complete technology platform (Time to Market) that covers the entire process from sending NIR + Lab data, to NIR Calibration as a Service, from online purchase of calibrations, to NIR Predictor software that directly evaluates newly measured NIR data locally and generates result reports.
Besides the free desktop version with user interface, the NIR Predictor can also be integrated (OEM). This can be integrated in parallel as a complement to your current Predictor, allowing the user to choose how they want to calibrate.
And give them the advantage in NIR feasibility studies and NIR spectrometer evaluations to quickly provide the customer with a solid and accurate calibration that will make their NIR system deliver better results.
Advantages for your NIR users (internal or external)
no initial costs (no chemometrics software license required),
calculable operating costs (fixed amount instead of time and hourly rate) (calibration development, calibration maintenance)
easy to use (no chemometrics and software training),
quicker to use (no calibration development work) and
Our chargeable service is based on the calibration development and the annual calibration use.
Calibration development and calibration use can also be carried out separately (manufacturer / user).
For you as a spectrometer manufacturer, this means that you can deliver your system pre-calibrated for certain applications without incurring software license costs. And without your application specialists having to provide additional calibration services.
The unique advantages of our calibration service together with the free NIR Predictor are:
no software license costs (chemometrics software, predictor software, OEM integration)
no chemometrics know-how necessary
no time needed to develop optimal NIR calibrations.
About CalibrationModel.com : Time and knowledge intensive creation and optimization of chemometric evaluation methods for spectrometers as a service to enable more accurate analysis and measurement results.
Using cost saving NIR-Spectroscopy Analysis? You can Save even more Costs and Time! How? Read here | VIS NIR NIRS Spectroscopy LabManager Labs QualityControl CostSaving foodindustry foodproduct Spectrometer Sensor Analytics LINK
Machine Learning for NIR Spectroscopy as a Service, a Game Changer for Productivity and Accuracy/Precision! Use the free NIR-Predcitor software to combine NIRS + Lab data and send your Calibration Request. LabManager Analysis MachineLearning LINK
Spectroscopy and Chemometrics News Weekly 24, 2020 | NIRS NIR Spectroscopy MachineLearning Spectrometer Spectrometric Analytical Chemistry Chemical Analysis Lab Labs Laboratories Laboratory Software IoT Sensors QA QC Testing Quality LINK
“Near-Infrared (NIR) Spectroscopy to Differentiate Longissimus thoracis et lumborum (LTL) Muscles of Game Species” LINK
“Estimation of Harumanis (Mangifera indica L.) Sweetness using Near-Infrared (NIR) Spectroscopy” LINK
“Handheld Near-Infrared Spectrometers: Reality and Empty Promises” miniaturization NIRS FTNIR MEMS MOEMS LVFs LINK
BESTCentreLTU research hot off the press: | In collaboration with Assoc. Prof. Kellie Tuck from , we’ve developed new near-infrared emissive electrochemiluminophores for sensing in NIR transparent biological media. LINK
“Near-Infrared Emitter for Bioanalytical Applications” NIR ECL electrochemiluminescence LINK
“Fault detection with moving window PCA using NIRS spectra for the monitoring of anaerobic digestion process” LINK
“New applications of visnir spectroscopy for the prediction of soil properties” LINK
“Simultaneous determination of quality parameters in yerba mate (Ilex paraguariensis) samples by application of near-infrared (NIR) spectroscopy and partial least …” LINK
“Control of ascorbic acid in fortified powdered soft drinks using near-infrared spectroscopy (NIRS) and multivariate analysis.” LINK
Infrared Spectroscopy (IR) and Near-Infrared Spectroscopy (NIR)
“Non-Invasive Blood Glucose Monitoring using Near-Infrared Spectroscopy based on Internet of Things using Machine Learning” LINK
“Investigating the Quality of Antimalarial Generic Medicines Using Portable Near-Infrared Spectroscopy” LINK
“Rapid quantitative detection of mineral oil contamination in vegetable oil by near-infrared spectroscopy” LINK
“THE DETERMINATION OF FATTY ACIDS IN CHEESES OF VARIABLE COMPOSITION (COW, EWE’S, AND GOAT) BY MEANS OF NEAR INFRARED SPECTROSCOPY” LINK
“Detection of melamine and sucrose as adulterants in milk powder using near-infrared spectroscopy with DD-SIMCA as one-class classifier and MCR-ALS as a means to provide pure profiles of milk and of both adulterants with forensic evidence” LINK
“Protein, weight, and oil prediction by singleseed nearinfrared spectroscopy for selection of seed quality and yield traits in pea (Pisum sativum)” LINK
“Modeling bending strength of oil-heat-treated wood by near-infrared spectroscopy” LINK
“ripening stages monitoring of Lamuyo pepper using a new‐generation near‐infrared spectroscopy sensor” LINK
“Should the Past Define the Future of Interpretation of Infrared and Raman Spectra?” LINK
“Performance Improvement of Near-Infrared Spectroscopy-Based Brain-Computer Interface Using Regularized Linear Discriminant Analysis Ensemble Classifier Based on Bootstrap Aggregating.” LINK
“Continuously measurement of the dry matter content using near-infrared spectroscopy” LINK
“Rapid identification of Lilium species and polysaccharide contents based on near infrared spectroscopy and weighted partial least square method.” LINK
“A Rapid and Nondestructive Approach for the Classification of Different-Age Citri Reticulatae Pericarpium Using Portable Near Infrared Spectroscopy.” LINK
Hyperspectral Imaging (HSI)
“Rapid detection of quality index of postharvest fresh tea leaves using hyperspectral imaging” LINK
“Non-Destructive Detection of Tea Leaf Chlorophyll Content Using Hyperspectral Reflectance and Machine Learning Algorithms” LINK
“A deep learning based regression method on hyperspectral data for rapid prediction of cadmium residue in lettuce leaves” LINK
“Deep learning applied to hyperspectral endoscopy for online spectral classification” DOI:10.1038/s41598-020-60574-6 LINK
“Detection of fish fillet substitution and mislabeling using multimode hyperspectral imaging techniques” LINK
Chemometrics and Machine Learning
“Molecules, Vol. 25, Pages 1453: Characterization, Quantification and Quality Assessment of Avocado (Persea americana Mill.) Oils” LINK
“Comprehensive Chemometrics – Chemical and Biochemical Data Analysis Reference Work • 2nd Edition • 2020” | books Chemometrics DataAnalysis Chemical Biochemical LINK
“Identification of invisible biological traces in forensic evidences by hyperspectral NIR imaging combined with chemometrics” LINK
Research on Spectroscopy
“Automatisierte und digitale Dokumentation der Applikation organischer Düngemittel” LINK
“Plenary Lecture Methods and Tools for Sensors Information Processing” LINK
Equipment for Spectroscopy
Using NIR scanner to assess grass in sward for composition prior to baling and wrapping for EU LIFE Farm4More project. Thanks to Dinamica Generale for providing the equipment LINK
“Determination of soluble solids content in Prunus avium by Vis/NIR equipment using linear and non-linear regression methods” LINK
“Characterization of Deep Green Infection in Tobacco Leaves Using a Hand-Held Digital Light Projection Based Near-Infrared Spectrometer and an Extreme Learning …” LINK
Agriculture NIR-Spectroscopy Usage
“Hyperspectral imaging using multivariate analysis for simulation and prediction of agricultural crops in Ningxia, China” LINK
“Placing Soil Information in the Hands of Farmers” LINK
“Robustness of visible/near and midinfrared spectroscopic models to changes in the quantity and quality of crop residues in soil” LINK
“Complex Food Recognition using Hyper-Spectral Imagery” LINK
Horticulture NIR-Spectroscopy Applications
” The Effect of Spent Mushroom Substrate and Cow Slurry on Sugar Content and Digestibility of Alfalfa Grass Mixtures” LINK
Laboratory and NIR-Spectroscopy
“The influence analysis of reflectance anisotropy of canopy on the prediction accuracy of Cu stress based on laboratory multi-directional measurement” LINK
“Klasifikasi Kopi Green Beans Arabika Sumatera Utara Menggunakan FT-Nirs dan Analisis Diskriminan” LINK
“APLIKASI NEAR INFRARED SPECTROSCOPY (NIRS) UNTUK MENGETAHUI KANDUNGAN HARA NITROGEN FOSFOR DAN KALIUM PADA INSTALASI …” LINK
” Identification of common wood species in northeast China using Vis/NIR spectroscopy” LINK
“Efficient Super Broadband NIR Ca2LuZr2Al3O12:Cr3+,Yb3+ Garnet Phosphor for pc‐LED Light Source toward NIR Spectroscopy Applications” LINK
“Performance comparison of sampling designs for quality and safety control of raw materials in bulk: a simulation study based on NIR spectral data and geostatistical …” LINK
“Prediction of drug dissolution from Toremifene 80 mg tablets using NIR spectroscopy” LINK
“Nearinfrared spectroscopy (NIRS) for taxonomic entomology: A brief review” LINK
“Determination of tomato quality attributes using portable NIR-sensors” LINK
“Exploring the potential of NIR hyperspectral imaging for automated quantification of rind amount in grated Parmigiano Reggiano cheese” LINK
“Rapid detection of adulteration of minced beef using Vis/NIR reflectance spectroscopy with multivariate methods.” LINK
“Evaluation of quinclorac toxicity and alleviation by salicylic acid in rice seedlings using ground-based visible/near-infrared hyperspectral imaging.” LINK
“An optimized non-invasive glucose sensing based on scattering and absorption separating using near-infrared spectroscopy” LINK
“Identification of waxy cassava genotypes using fourier‐transform near‐infrared spectroscopy” LINK
“Kernel functions embedded in support vector machine learning models for rapid water pollution assessment via near-infrared spectroscopy” LINK
“Near-infrared-based Identification of Walnut Oil Authenticity” LINK
“Detection of flaxseed oil multiple adulteration by near-infrared spectroscopy and nonlinear one class partial least squares discriminant analysis” LINK
“Application research of sensor output digitization for compact near infrared IOT node” LINK
“Refining Transfer Set in Calibration Transfer of Near Infrared Spectra by Backward Refinement of Samples” LINK
“Highly identification of keemun black tea rank based on cognitive spectroscopy: Near infrared spectroscopy combined with feature variable selection” LINK
“Optimizing Rice Near-Infrared Models Using Fractional Order SavitzkyGolay Derivation (FOSGD) Combined with Competitive Adaptive Reweighted Sampling (CARS)” LINK
“Fourier transform near infrared spectroscopy as a tool to discriminate olive wastes: The case of monocultivar pomaces.” LINK
“Evaluation of quinclorac toxicity and alleviation by salicylic acid in rice seedlings using ground-based visible/near-infrared hyperspectral imaging” LINK
“Near Infrared Spectrometric Investigations on the behaviour of Lactate.” LINK
“Nondestructive rapid and quantitative analysis for the curing process of silicone resin by nearinfrared spectra” LINK
“An introduction to handheld infrared and Raman instrumentation” LINK
“Hyperspectral anomaly detection by local joint subspace process and support vector machine” LINK
“Assessment of matcha sensory quality using hyperspectral microscope imaging technology” LINK
“Application of Infrared Spectroscopy and Chemometrics to the Cocoa Industry for Fast Composition Analysis and Fraud Detection” LINK
“Calibration models for the nutritional quality of fresh pastures by nearinfrared reflectance spectroscopy” LINK
“Achieving robustness to temperature change of a NIRS-PLSR model for intact mango fruit dry matter content” LINK
“Application of hyperspectral imaging combined with chemometrics for the non-destructive evaluation of the quality of fruit in postharvest” LINK
“Characterization of Deep Green Infection in Tobacco Leaves Using a Hand-Held Digital Light Projection Based Near-Infrared Spectrometer and an Extreme Learning Machine Algorithm” LINK
“MEMS technology for fabricating plasmonic near-infrared spectrometers” LINK
“Sensors, Vol. 20, Pages 545: Development of Low-Cost Portable Spectrometers for Detection of Wood Defects” LINK
“Classification of Granite Soils and Prediction of Soil Water Content Using Hyperspectral Visible and Near-Infrared Imaging” LINK
“Determining mandatory nutritional parameters for Iberian meat products using a new method based on near infra-red reflectance spectroscopy and data mining” LINK
“Instrumental Procedures for the Evaluation of Juiciness in Peach and Nectarine Cultivars for Fresh Consumption” LINK
“The creation of the FT-NIR calibration for the determination of the amount of corn grain in concentrated feed” LINK
In this 9th clip from his presentation at the 2019 IFS Agronomic Conference, Wouter Saeys explains which type of NIR is best for measuring the nutrient content of manure, and why. Info on this paper is here; it’s free for Society Members to download: LINK
“Remote Sensing, Vol. 12, Pages 928: Machine Learning Algorithms to Predict Forage Nutritive Value of In Situ Perennial Ryegrass Plants Using Hyperspectral Canopy Reflectance Data” LINK
“Development of a Method To Prioritize Protein-Ligand Pairs on Beads Using Protein Conjugated to a Near-IR Dye.” LINK
“Agronomy, Vol. 10, Pages 148: Performance Evaluation of Two Commercially Available Portable Spectrometers to Non-Invasively Determine Table Grape and Peach Quality Attributes” LINK
“Investigation of a Medical Plant for Hepatic Diseases with Secoiridoids Using HPLC and FT-IR Spectroscopy for a Case of Gentiana rigescens” LINK
Food & Feed
“Comparison of sensory evaluation techniques for Hungarian wines” LINK
“Roadmap of cocoa quality and authenticity control in the industry: A review of conventional and alternative methods” LINK
We build the optimal quantitative prediction models for your NIR analytical needs (No need for mathematical/statistical model building software usage at your site).
The NIR-Predictor software and the calibration models are at your site. No internet connection needed to our service. You can do unlimited predictions, that allows fast measurement cycles with no extra cost (Not payed per prediction).
You own your NIR + Lab data and the calibrations. You can have access to the detailed Calibration Report with all the settings and statistics (No Black-Box Models).
Calibration Model simplifies the process of training machine learning models for NIRS data
while providing an opportunity to trying different algorithms and applied near-infrared spectroscopy (NIRS) knowledge.
It’s more than an AutoML platform, it’s a full service where you can download the optimal model and its describing Calibration Report
that provide insights into the data preparation, feature engineering, model training, and hyperparameter tuning.
Open the demo Spectra folder by using the Menu > Open Demo Spectra or press F8.
There are files with spectra from different Vendors.
Drag & drop a spectra file onto the NIR-Predictor window (or press Ctrl+O as for ’Open some files).
The spectra will be
All the steps are fully automatic.
All calibrations that are compatible with the spectra, will produce prediction results in one go.
To select specific calibrations choose the Application. Where the ” ” empty means use all the calibrations.
To define a Application read more in chapter “Applications”
To get access to Statistics of Predictions and Reports use the Menu > Show more/less (Ctrl+M) or you can simply resize the window. Here you can also re-do the Analyze step manually with changed inputs (e.g. Result Ordering).
Creating your own Calibrations
How it works – step by step
You have measured your samples with you NIR-Instrument Software.
And got the Lab-values of these samples.
Note: If you combined these data already in your NIR software used,
and you can export it as a JCAMP-DX file then use
Menu > Create Request File .req ... (F2)
and read the “Help.html” and NIR-Predictor JCAMP.
Else proceed as below.
The NIR-Predictor provides tooling for that:
Menu > Create Properties File... (F6)
Select the folder with your NIR spectra measured for an application.
NIR-Predictor creates a customized Properties file template for that data to enter the Lab values.
Note: You don’t need to specify your instrument or vendor or an application. It’s all done automatically. And also the sample spectra are detected and grouped automatically!
Use your favorite editor or spreadsheet program to enter and copy&paste
the Lab-references Values into the columns “Prop1”, “Prop2” etc. and save the file.
A final check of your entered data is done by NIR-Predictor,
to make sure your data ist complete and all is fine.
Menu > Create Calibration Request... (F7)
Select the folder with the filled file.
A CalibrationRequest.zip is created with the necessary data
if enougth diverse Lab values are entered.
Email the CalibrationRequest.zip file
to info@CalibrationModel.com to develop the calibrations.
When your calibrations are ready, you will receive an email with a link
to the CalibrationModel WebShop where
you can purchase and download the calibration files,
that work with our free NIR-Predictor software without internet access.
Note: Your sent NIR data is deleted after processing.
We do not collect your NIR data!
Note: Further details can be found under “Create Properties File” and “Create Calibration Request”.
Configure the Calibrations for prediction usage
in NIR-Predictor : Menu > Open Calibrations (F9)
an explorer window is opened where the calibrations are located
create a folder for your application, choose a name
copy the calibration file(s) (*.cm) into that folder
in NIR-Predictor : Menu > Search and load Applications (F4)
in NIR-Predictor : open the Application drop down list, and select your application by name
if all is fine, the calibration file is valid and not expired, it shows : Calibration “1 valid calibation”
the NIR-Predictor is now ready to predict
to switch the application, goto 6.
The Application concept allows to group multiple Calibrations together for an Application. By selecting an Application before prediction, only the Calibrations belonging to the Application will be used for Prediction. In the Demo Data this is used to have multiple spectrometer as Application. This can be used easily as e.g. as Application “Meat Products” containing Fat and Moisture Calibration.
To create an Application, create a folder with the Application’s name inside the Calibrations folder, and move/copy all the Calibrations files to this Application folder. To remove a Calibration from the Application, remove the Calibration file from the Application folder.
After creating an new Application folder, press menu Search and load Applications (F4) to update the NIR-Predictor dialog where the Application can be selected via the dropdown list. You don’t need to close the NIR-Predictor.
After moving Calibration files around, press menu Search and load Calibrations (F5) to update the NIR-Predictor dialog.
The use-all case
In the NIR-Predictor dialog where the Application can be selected via the dropdown list, the empty "" name means that all (yes all) valid Calibrations will be used for prediction.
Note: The Prediction Report will contain only results from spectral compatible Calibrations with the given spectra. That allows to automatically handle the multi vendor NIR instrument usage.
Prediction Result Report
Histograms of Prediction Values per Property
Shows the distribution of the predicted results per calibration. The histogram range contains the range of the calibrated property and includes the predicted results.
The histogram bar (bin) color is defined as follow:
blue : all predictions inside calibration range.
red : all predictions outside calibration range.
orange : some overlaps with calibration range.
So not all spectra in a orange bin are outside calibration range.
Note: Predicted values are always shown in Histogram table and Prediction Value List table, even if the spectrum does not fit into model (spectrum different to model, aka Residual Outlier) shown as Out = X.
Note: Old browsers like Microsoft Internet Explorer 11 don’t support the grafics for Histogram charts. Use an current browser like Firefox or Chrome or Edge.
Note: If your browser opens the report too slow, try to deactivate some browser plugins, because they can filter what you look at and some add-ons are really slow.
Spectra Plot Thumbnail on the Prediction Report
Visualizes the min,median,max spectrum of the spectra dropped as files on the NIR-Predictor. This gives a minimal and good spectral overview of the predicted property results.
Spectra Plot color legend: min,median,max spectrum by predicted property or if no calibration is available by spectral intensity.
The min,median,max is determined from the predicted properties or if not available from the intensity of the spectra.
Beside the histogram of the predicted properties, where the distribution can be seen, the spectra shown are the ones from min,median,max predicted property.
This gives a minimal and good spectral overview of the predicted property results.
The “Spectral Range” and number of datapoints is shown in the Prediction Report Header below the listed spectra files.
To zoom the spectra plot a little, zoom the report in the browser (hold ctrl + mouse wheel, or pinch on touch screen).
The spectra plots and histograms are stored with the report and can be archived.
Note that the spectra are shown in the raw values that are loaded, they are not shown pre-processed as the calibration model uses them to make the prediction.
Note that the median property spectrum is the median from the predicted property pobulation and not the “median” of the calibration property range.
Note that in the multi calibration prediction case, the spectra are selected for each property based on the related predicted property values and so the spectra plots shows typical different spectra.
To safeguard the prediction results, outliers are automatically checked for each individual prediction. This is based on limits that are determined when creating the calibration with the base data. Thus, a strange spectral measurement can be detected and signaled as an outlier even without base data only by means of the calibration and the NIR predictor. A prediction result with outlier warning is to be distrusted. How the various outlier tests are interpreted and how to avoid them in practice is described here.
The spectrum is an outlier to the model, if the spectrum is not similar with the spectra and lab-values the model is built with.
This legend is shown on each NIR-Predictor prediction report below the results:
Outlier (Out) Symbol Description
“X” : spectrum does not fit into model (spectrum different to model)
“O” : spectrum is wide outside model center (spectrum similar to model but far away)
“=” : prediction is outside upper or lower range of model (property outside model range)
“-” : spectrum is incompatible to calibration
Note: A prediction result with outlier warning is to be distrusted.
There are 3 outlier cases (X, O, =) and the incompatible data case “-”.
The bad case is “X”
the medium case is “O”
and the soft case is “=”.
The technical names in literature correspond to:
“X” : Spectral Residual Outlier
“O” : Leverage Outlier
“=” : Property Range Outlier
These 3 outlier cases can appear in combinations, like “XO=” or “XO” or “O=” or “X=”. The more outlier marker are shown the more likely the spectrum is an Outlier.
The default setting in NIR-Predictor Menu > “Report with Simplified Outlier Symbols”
is ON, that will show only the worst case instead of all combinations to have a simplified minimal information.
if OFF, that will show the combinations (e.g. “XO=” or “XO” or “O=” or “X=”), which is more informative for analyzing problem cases.
Some hints to avoid these Outliers:
“X” : spectrum does not fit into model (spectrum different to model)
Check if the spectrum is noise only, or has no proper signal. That can happen when measured past the sample or measured into the air or at a different substance. If you have multiple NIR instruments of the same type, use spectra measured with different instruments for the calibration.
“O” : spectrum is wide outside model center (spectrum similar to model but far away) Sample temperature has an effect on NIR spectra shape, use spectra measured at different (typical use) temperatures (sample temperature, instrument temperature).
“=” : prediction is outside upper or lower range of model (property outside model range)
Use more spectra for the calibration in the Lab value region where your special interest is. If the predicted value is only a little bit out of the calibration range, it can be Ok. Add these spectra to the calibration spectra (with the Lab values), to extend the prediction range of the calibration.
“-” : spectrum is incompatible to calibration
The spectra (from the NIR instrument) has a different wavelength range or a different resolution than the spectra used for calibration. Check Instrument settings (wavelength range, resolution)
To change the ordering, a drop-down-box is located below the Analyze button. If there is an analysis from the current session, and the Result Ordering is changed, the data is re-Analyzed and reported with the new Result Ordering setting. That allows to compare the different orderings. The Result Ordering is listed in the Prediction Report above the Prediction Value List and stored in the settings.
The order/sorting of the prediction results of the spectra can be defined:
GivenOrder (default) the given order of the spectra from file select dialog or drag&drop
*) sorted : ascending sort
Date_Name sorted by Date (if any) and then by Name
Name_Date sorted by Name and then by Date
Date_NamesWithNumbers sorted by Date (if any) and then by Name with number logic
NamesWithNumbers_Date sorted by Name with number logic (e.g. “ABC1” is before “ABC002” ) and then by Date
*) as above but sorted Rev : reverse sort = descending sort
E.g. with reverse sort by Rev_Date_Name, the newest spectra appear on top.
Print to PDF
Depending on how many calibrations are used the result table is getting broader. To print the report (e.g. to Adobe PDF, FreePDF or Microsoft XPS), sometimes the landscape format is shorter in number of pages or in portrait a scale of 80% fits nicely. Or try another internet browser (Mozilla Firefox, Google Chrome, Microsoft Edge, …) to print the report and set the browser as your default browser so it will be opened by default.
Each report is contained in one file only, including the grafics. To save storage space the report file folder can be compressed to a zip file (.zip, .7z).
Enter lab values to NIR spectra
Entering the laboratory reference values for NIR calibrations
We have developed specialized tools into NIR-Predictor to combine the NIR and Lab data is a sample-based safe manner.
The main target is to improve Data Quality during the step of combining of the Lab data and the NIR data, because to model a good reliable calibration the data that build the base needs to be of high quality.
It also simplifies to enter the lab values manually to the corresponding NIR data, because of automatically grouping repeated NIR measurements of the same sample, so the lab values can be entered sample based and not by spectrum.
It helps to avoid false reference data, because of the broken relation of NIR spectra and reference values, data entry on the wrong position in the table.
And Helps to detect errors of duplicated or multiple copies of spectra files, and checks for inconsistencies in Date-Time and Sample-Naming. It also checks for missing values.
That all increases the Data Quality for the next step of Calibration Development, and makes data entry a less time consuming and less risky work.
How it works
Menu > Create Properties File... (F6) select the folder with your NIR spectra measured for an application. NIR-Predictor creates a Properties file template for that data : PropertiesBySamples.csv.txt
Use your favorite editor or spreadsheet program to enter and copy&paste the Lab Values into the columns and save the file.
Menu > Create Calibration Request... (F7) select the folder with the filled file for a last check and a Calibration Request file is created with the needed data as a single zip file.
Email the Calibration Request file to info@CalibrationModel.com to develop the calibrations.
Ok that is it, the NIR-Predictor guides you through the steps needed. And if you need to know more details, the Chapter “Create Properties File” is for you.
Create Properties File
If you have (exported) JCAMP-DX files containing the Lab-Values, you don’t need to do this step.
You can send the JCAMP file with your Request (.req) file directly to the calibration service at info@CalibrationModel.com.
If your JCAMP-DX files does NOT contain Lab-Values, this is a way to go.
For calibrating the spectra to the lab-values you need to assign the lab-values to the spectra. The easiest way is to have a table where each spectrum (row) is linked to multiple lab-values (columns). This function Create Properties File build such a table for the selected spectra folder automatically!
This table is stored in the file PropertiesBySamples.csv.txt. This can be created for any spectra folder you like. The file extension is .csv.txt to make it easy to edit in a text editor and also in a spreadsheet (excel). The columns are standard TAB separated.
Prop1, Prop2, Prop3 are the place to enter the Lab Reference Concentrations properties corresponding to each spectrum. It can be extended to Prop4, Prop5, … etc. Of course you can enter real word names like “Fat (%)” instead of “Prop1”. It’s recommended to put the measurement unit beside the name.
Replicates is the number of replicated or repeated spectra of a sample that is grouped together in the Sample based property file. Sample name and the DateFirst / DateLast between the sample spectra are measured.
Date format is ISO-8601. Missing Dates are 0002-02-02T00:00:00.0000000.
If the file PropertiesBySamples.csv.txt already exist in the selected folder, the user will be notified (it will not be overwritten, because the file may contain user entered Lab-values). The Lab Reference Concentrations values are initialized to 0 (zero) and needed to be changed.
Note: 0 is not interpreted as missing value! If you have a 0 concentration value, put in 0 or 0.0 .
The entry of properties is as easy as possible, because it’s organized by Sample (and not by Spectra), so it’s like your Lab-Value Table that is sample based. The sample rows are sorted in a special way by Sample name. Sorting by Date or alphabetically by Sample can done easily in a spreadsheet program.
Note: when coping lab values to the samples make sure they correspond, so that there are no gaps and the sorting is the same.
The Spectra (rows) are initially sorted by name (and date) to have the replicates/repeats together. You can sort for your convenience in a spreadsheet program.
Enter the Lab Reference Concentrations to the spectra/sample.
Enter the Lab-Values in spreadsheet (e.g. Excel) or a text editor (e.g. Notepad++). If done, use the next menu Create Calibration Request.
Hints: Data handling:
The NIR-Predictor creates the PropertiesBySamples.csv.txt once, after that the user is responsible for its content. That means NIR-Predictor does not change this file anymore.
You can remove entire rows (spectra) in the property file. You don’t need to remove the spectra files. The NIR-Predictor is aware of the relation, the PropertiesBySamples.csv.txt defines what will be calibrated.
How to add more spectra files?
The additional spectra can be handled in a separate folder, create the property file and copy the spectra to the other folder and copy/merge the property files together in your editor or spreadsheet.
Copy the spectra into the folder, rename the PropertiesBySamples.csv.txt to e.g. “PropertiesBySamples-Part1.csv.txt” and use Create Properties File to create a new PropertiesBySamples.csv.txt with all the spectra. You can copy/merge the content of the Properties files together in your editor or spreadsheet.
What happens with possible duplicate rows? It does no harm to the Calibration because we do an exact checking and data cleaning in the calibration process.
What happens to duplicate spectra names? The spectra names are not relevant for the calibration process. The spectra names are helpful to assign the lab-values to the corresponding spectrum entry. That’s why the table is initially sorted by name. The spectra names can be edited by the user.
Create Calibration Request
The menu function Create Calibration Request packs a created Properties file (see “Create Properties File”) and it’s linked spectra files in a compressed ZIP file for sending to the CalibrationModel.com Service.
Please note that the number of measured quantitative samples need to be at least 60 . That means you need at least 60 different spectra (not counting the replicate/repeated measurements).
It shows additional property information about the data you have entered, like – the property type (Quantitative) – it’s range (min – max) and – the number of unique values and – if the Lab-values are enough diverse to get calibrated.
First select the folder with the PropertiesBySamples.csv.txt and measured spectra files of samples you have Lab-values. The data is checked and you get notified what is missing or might be wrong. If something needs to be changed, edit the PropertiesBySamples.csv.txt and do Create Calibration Request again. Your last selected folder is remembered, so you can press return in the folder selection dialog.
Hint: The keyboard shortcuts for redoing it after you edited some entries is : F7 Return – that allows you to get the property information quickly.
Hint: If you open the PropertiesBySamples.csv.txt in a spreadsheet program, you can create Histogram plots of the entered Lab-values, to see in which range are to less samples measurements.
When all is fine
When all is fine the “CalibrationRequest.zip” file is created for that data.
The ZIP file contains:
your personal REQuest file for your computer system, that looks like
e.g. “337dcdc06b2d6dfb0b5c4bba578642312edf2ae84d909281624d7e26283e8b07 WIN-GB0PB48GSK4.req”
the spectra data files
Note: If the CalibrationRequest.zip file is already created and you change the PropertiesBySamples.csv.txt make sure to delete the old CalibrationRequest.zip file first! In the dialog it states if it was successfully created or NOT because it already exist. So you are always on the safe side.
Note:CalibrationRequest.zip file name contains the property names to know what would be calibrated and at the end an identification number for referencing the file. E.g. “CalibrationRequest ‘Prop1’ – ‘Prop2’ h31T3wOH.zip”
The users program settings are stored in UserSettings.json
The program counters are stored in GlobalCounters.json