Unscrambler的预测含量论述.ppt

Data Analysis using The Unscrambler? Changchun Institute of Applied Chemistry  Ranjit Viswanathan Manager – Sales (Asia Pacific) CAMO Software India Pvt Ltd ranjitv@ # +91 98456 98425 Background Changchung has given two sets of Files JCAMP-DX files, assumed to be Spectra readings. We will consider these as “Input” (X) variables. Excel Sheet, with Protein% and Fat% of Soybeans We will consider these as “Output” (Y) variables. There is one extra JCAMP-DX reading of S74, which does not have Protein/Fat values. We will treat this as “Unknown Sample”, where Protein% and Fat% have to be predicted, using The Unscrambler. The objective is to Predict the Protein/Fat% values of S74. Overall Steps that we are following Ensure Data Consistency in JCAMP-DX and Excel Import data into The Unscrambler Setting up the data in The Unscrambler for analysis Preliminary Analysis using The Unscrambler Create a Regression Model Improve Regression Model Save Regression Model Prediction using The Unscrambler More options for Prediction Ensure Data Consistency in JCAMP-DX and Excel Rename some files in Windows Explorer Since we are importing data from two different sources into The Unscrambler, it is important to ensure that all samples/variables are in the same order in both sources. Rename files “S1_0.dx” to “S01_0.dx” This ensures that all the files will be listed in Ascending order in The Unscrambler. Similarly, rename “S2_0.dx”, “S3_0.dx” “S4_0.dx” and “S6_0.dx” by adding a 0 before the digit. Result of renaming The order of the files, AFTER renaming. Analyze Excel File There should be no mismatch in the data from the two sources. Therefore, for all the IR readings, please ensure that there are Protein/Fat% values available. Currently, there is a mismatch in the data. In Excel, the sample S69 does not have a JCAMP reading. Delete this sample from Excel. In Explorer, there is additional reading of S74, which is not there in Excel. Add S74 sample name, with blank values of Pr