wine_variety

---

1.About

• a predictive model to identify wine variety with 10 feature

---

---

2.Dataset

• you can download the dataset from->https://drive.google.com/file/d/1kq8PSsj1U_naNzEsUTnmumWy77L8SKI5/view

---

---

3.Accuracy

• this model has 58.8 % accuracy on training set

---

---

4.Data Preprocessing

• fill the null value in the dataset
• combine region_1,region_2 and province to region
• encode the region,country and winery
• reduce the range of point
• preprocess the designation and review_description

i have save the processed data into processed_data.csv file

---

---

5. Other Files

• processed_data.csv ->i have saved the processed data into this file.so you dont have to preprocess it
• wine_model.pkl ->i have saved the xgboost trained model into this file.so you donot have to train it
• sample_submission1.csv->this is the submission file where the predicted value of test data is stored

---

---

6.Model Used

• i have used xgboost model to predict the variety .it has 59 % accuracy in train dataset.

XGBOOST-->

Accuracy Score->58.8%

Confusion Matrix->

Feature Importance->

---