Conclusion and Scope - Ashok-Ab4/House-Prices-AirBnB GitHub Wiki

Results and Performance :

The models were applied to the given dataset and the performance metrics are as shown above. This was then plotted using 2 plots to find the best performing model. Screen Shot 2021-01-19 at 10 08 59 PM

The RF model was chosen as the best performing model and was then used to make predictions on the test dataset which was then compared to the actual test price. The summary of that was as follows:

Screen Shot 2021-01-19 at 10 09 23 PM

The prediction was overall permissible but had wide variations as displayed by the minimum and maximum values.

Discussion and Recommendations:

  1. Take time to go through the data in depth and trying to find the underlying connections between variables that weren’t too readily visible in this particular iteration of model building
  2. Take more time to properly analyze the non linearity of the data that couldn’t be explored in depth due to time constraints
  3. Perhaps make more improvements to the data because there were clearly no correlations between the response and the predictors so making an accurate model with the current state of the data is a tough ask.
  4. Do cross validation and resampling more thoroughly and use a bigger chunk of the data. None of these could be explored in this project due to computational deficiencies of the system.