To use the Automobile Price Prediction:
-
Open
main.py
and edit the jsondata
with the automobile characteristics for which price is to be predicted -
For batch input, you can use
main_batch.py
- The raw data contains a lot of missing data. So, those columns in which a lot of data is missing is skipped. Even if a single data element is missing, we remove that whole row.
- The features are extracted whose impact on price is observable, and which are not are discarded.
- Our labels are
price
, and on the extracted data we do a 70% split as training data, and other 30% as testing data. This splitting is done on randomly shuffled data. - I have modelled this predicitive experiment in
Microsoft Azure Machine Learning Studio
usingLinear Regression
as initialising model. - By feeding this Regression algorithm along with the training data into a model which will be trained according to the features and behaviour in training data.
- Now using this newly trained model and the testing data available, we make predictions for the testing data and stored in
scored labels
column and store it in a csv file. - According to these predictions, we compute the coefficient of determination and mean error values.