Wine Quality Prediction

The data file wine_train.csv contains a dataset of 3673 data points with 11 features and a label which ranges from 3-9.The aim of the project is to predict the quality of the wine by training the data on the given data points and then creating a separate csv file for testing and feedinf the predicted value of quality in a new column and marking it as good, average or bad. Any value of quality predicted which is less than 6 is marked as 'bad', equal to 6 marked as 'average' and greater than 6 is marked as'good'.Out of 3673 data points, 3000 are used to train the model and the rest are used for testing.

Technology used- Python3, Random forest, Hyperparameter tuning.

Data Description

fixed.acidity
volatile.acidity	
citric.acid	
residual.sugar	
chlorides	
free.sulfur.dioxide	
total.sulfur.dioxide	
density	
pH	
sulphates	
alcohol	
quality (label)

The final prediction is based on Random Forest Regressor and dumped into Testing and prediction.csv file. Train accuracy is around 0.96 which shows that it has low bias.