In data exploration and modeling, we utilize all features to predict the quality of wine. We perform deep learning sequential method to achieve 77.3 % accuracy rate.
There are two possible predicted classes: "Fair" and "Very Good". If we were predicting the quality of wine, "Very Good" would mean to have better quality, and "Fair" would mean to lower quality of wine.
The classifier made a total of 357 predictions.
Out of those 357 cases, the classifier predicted "Very Good" 2 times, and "Fair" 355 times.
In reality, 77 cases in the sample are in category of very good quality, 280 cases are not.
Top Five Features
Summary:
In data exploration and modeling, we perform importance features methods to select the top five features to predict the quality of the wine. We perform a deep learning sequential method to achieve an 81.8 % accuracy rate. In comparing to the result with all features given, we are able to improve the accuracy rate by eliminating the unnecessary features.
There are two possible predicted classes: "Fair" and "Very Good". If we were predicting the quality of wine, "Very Good" would mean to have better quality, and "Fair" would mean to lower quality of wine.
The classifier made a total of 357 predictions.
Out of those 357 cases, the classifier predicted "Very Good" 13 times, and "Fair" 344 times.
In reality, 67 cases in the sample are in category of very good quality, 290 cases are not.
Red Wine
All Features
Features by Score Ranking
alcohol - 0.171243
sulphates - 0.112875
volatile acidity - 0.09814
citric acid - 0.096145
density - 0.085803
total sulfur dioxide - 0.082951
fixed acidity- 0.075679
residual sugar - 0.073609
free sulfur dioxide - 0.069072
pH - 0.067486
chlorides - 0.066996
Target Value:
Quality - Fair and Very Good
Methods:
Training size : 0.7
Deep Learning Model: Sequential
Summary:
In data exploration and modeling, we utilize all features to predict the quality of wine. We perform deep learning sequential method to achieve 90.2% accuracy rate.
There are two possible predicted classes: "Fair" and "Very Good". If we were predicting the quality of wine, "Very Good" would mean to have better quality, and "Fair" would mean to lower quality of wine.
The classifier made a total of 123 predictions.
Out of those 123 cases, the classifier predicted "Very Good" 8 times, and "Fair" 115 times.
In reality, 15 cases in the sample are in category of very good quality, 108 cases are not.
Top Five Features
Summary:
In data exploration and modeling, we perform importanace features methods to select the top five features to predict the quality of wine. We perform a deep learning sequential method to achieve an 88.6 % accuracy rate. As a result, the accuracy rate has been slightly dropping by eliminating some of the features.
There are two possible predicted classes: "Fair" and "Very Good". If we were predicting the quality of wine, "Very Good" would mean to have better quality, and "Fair" would mean to lower quality of wine.
The classifier made a total of 123 predictions.
Out of those 123 cases, the classifier predicted "Very Good" 0 times, and "Fair" 123 times.
In reality, 18 cases in the sample are in category of very good quality, 105 cases are not.
