Clean, Scale & Split data to train Measure Mean Absolute/Square Error for DecisionTree, LinearRegression, RandomForest using cross validation to evaluate the best algorithm to model bike rental forecasting. RandomForest being the best performing is GridSearched to find the best model parameters. Then predict the test data output & compare with actual test data output to measure mean square error and plot the output. It was fun. Data description, metrics and plots in notebook cells inline.

Training various Models (Decision Trees, SGD, RF, Softmax, VotingClassifier ensemble, Cross validation for each model, Grid Search of best hyperparameters & training the model. MNIST contains images of 28x28 pixels each become feature to predict the name of object.