Pemilu Prediction Model #Assignment4

I am going to create prediction model of Pemilu using Orange. Orange is an open source for data mining processing. Compare to other data mining software, orange excels in visualization or we call visual programming.

Creating prediction model using data training on data pemilu using the following algorithm:

1. Decision Tree (C4.5)

2. Naive Bayes (NB)

3. K-Nearest Neighbor (K-NN)

Here are the steps of each algorithm:

1. DECISION TREE MODEL

STEP 1

Open orange application. It will show the first interface like the picture below.
Write the title as you want. You can write the description or leave it blank. Then click ok.

 

STEP 2

In creating decision tree, drag the icon file, data table, tree, and tree viewer to orange worksheet.

STEP 3

Chose file you want to use. Click reload, then apply.
Note: In column terpilih atau tidak (categorical, target)
In column nama calon legislative (categorical, meta)

STEP 4

Connect all the icons like the following picture.

We connect the icons to view the data as data table in tree model and visualize it in tree viewer.

STEP 5

Click icon tree viewer and we got the result as the following picture.

2. NAIVE BAYES

STEP 1

•         Drag the icons as the following picture
•     Connect the icons
•     We get the analysis in test score (AUC, CA, F1, Precision, Recall), Confusion Matrix, and ROC Analysis
• 
  

STEP 2

Click Test and Score to see the result as the following

STEP 3

Click Confusion Matrix to see the result as the following.

STEP 4

Click ROC analysis to see the result as the following.

3. K-NEAREST NEIGHBOR (KNN)

STEP 1

Drag the icon as the following picture, and connect them.

STEP 2

Click Test & Score to see the result about AUC, Precision, etc.

STEP 3

See the result in distributions visualization. We can see the graph grouping by: Terpilih atau Tidak, fold, and other more.

10 Cross Validation is used in Test & Score for sampling.