Diamonds Machine Learning | Project

Objective

Create the best regression model of machine learning that can predict the price (independent variable) of the diamonds when a certain characteristic (features) are given.

Instructions

Two datasets are given in a kaggle competition

• Train Set: with the following columns.

Feature	Description
id	only for test & sample submission files, id for prediction sample identification
carat	weight of the diamond
cut	quality of the cut (Fair, Good, Very Good, Premium, Ideal)
color	diamond colour, from J (worst) to D (best)
clarity	a measurement of how clear the diamond is (I1 (worst), SI2, SI1, VS2, VS1, VVS2, VVS1, IF (best))
x	length in mm
y	width in mm
z	depth in mm
depth	total depth percentage = z / mean(x, y) = 2 * z / (x + y) (43--79)
table	width of top of diamond relative to widest point (43--95)

Independent variable	Description
price	price in USD

• Test Set: With the same features of the training set but in this case without the price, that has to be predicted using the ML model.

• Submit Set: Id the test set with the only columns of “id” and “price”. This set has to be submitted to Kaggle competition that will return a score based on the MRSE. The lower this score, the better model you are delivering.

Resources and libraries

• Sklearn | Machine Learning Library with regression models
• TensorFlow Keras | ML library
• h2O | Auto ML module for machine learning

Inputs

• diamonds_train.CSV | Shape (40345, 11)
• diamonds_test.CSV | Shape (13449, 10)

Outputs

• Models submitted to the kaggle competition
• Jupyter Notebooks of the process of each model

Methodology

1. Cleaning Dataset The following decision has to be done

   • Ensure not Null values
   • Avoid not numeric values
   • Drop the high correlated features
   • Standardize or normalize high-value range features

2. Identify the best model Different models apport different results. Some of them are analysed in the project.

   • LinearRegression
   • RandomForestRegression
   • LassoCV
   • SGDRegressor
   • Keras Sequential Dense Neural Network
   • SVR
   • Polynomic RFR
   • Sum of two Polynomic RFR

3. Analyse the result There are metrics from sklearn that help you to choose the best model

   • R2_score
   • MRSE
   • Scatter plot y_test vs y_true

Results

As it can be appreciated, the best solution addressed is the Polynomic Random Forest Regressor. These are the plots that demmonstrate the relationship between the true and the predicted values

Plot

Scatter

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

Project status

Learning and enjoying every day. Next steps:

• Test new models and explore better neural networks
• Use h2O to learn the AutoML function