/wine-variety

It's a predictive model that uses classification model (Simple Vector Machine and Decision Tree Algorithm ) to predict the wine "variety". The outputs along with all features are provided to abc.csv (after training, automatically generated CSV file). *All the asked questions are answered in the README.md or simply scroll down.

Primary LanguageJupyter Notebook

wine-variety

Contents of the Project:-

  1. Model used
  2. Features extracted
  3. Model accuracy in train
  4. Data Visualization
  5. Top 5 actionable insights
  6. Saving the Output in a csv
  7. Licensing

Elaboration of the contents :-

1) Model used : SVM and Decision Tree (Classification)

Capture4 Capture5

2) Features extracted:(price, points, provinces)

sass

3) Model Accuracy in Train:(accuracy_score =0.72216,f1_score = 0.6627)

Capture6 Capture7

4) Data Visualization(Decision Tree):

VarTree

5) Top 5 actionable Insights from the Data:

a) We can determine the "Significant Factors" or "Dominating Factors" that affects the prediction and thus in a way the mindset of the customer groups.

b) Using this Classication groups, we could easily "group" or "cluster" similar customer together for better implications and operations.

c) By the visualization of data using Decision Tree , we could easily see the "Randomness" or "dissimilities bewteen customers in the same group and other groups as well.

d) We can easily determine the "popularity" of the variety among the customers and the reason for it as well.

e) The most crucial insight is "Increasing the net profit (prices)" depending on the popularity determined earlier and make tactics from buisness point of view.

6) Saving the Output in a csv:

After predicting is saved automatically in the test.csv file under the column name "Predicted Value":

Before running the script:

Capture1111

After running the script:

tempsnip

7)Conversion of variety to its unique counterpart:

Capture

License

The project is available as open source under the terms of the MIT License.