Fingerprint to Zone Classification

Used libraries

• pandas
• numpy
• matplotlib
• sklearn
• keras with tensorflow backend

Steps

• loading data from csv
• shuffle data
• as input data has dictionary of features, we need to transform it to feature columns
• scaling features to (0,1)
• as its multi-class classification problem, we need to use one-hot encoding to transform labels
• splitting data into train, validation and test data sets
• defining and training simple NN with 1 hidden layer of 10 neurons using keras
• defining and training 2 more NN (one with neurons in hidden layer and one with more layers)
• comparing performance of all 3 models
• providing results, and showing predictions of the best model