Reference no: EM132400011

CSCE 822 - Data Mining & Warehousing Assignment - University of South Carolina, USA

Attached melb_data.csv file is the Snapshot of Tony Pino's Melbourne Housing Dataset. Do the following data preprocessing and apply KNN and RandomForest algorithms to classify the property prices.

1. Fill the missing values in the dataset using imputation approaches as we talked in class. You can use the scikit-learn's module

from sklearn.impute import SimpleImputer

my_imputer = SimpleImputer()

data_with_imputed_values = my_imputer.fit_transform(original_data)

The default imputer use mean values to fill the missing values. You can try other imputation method as well.

2. Replace the categorical/nominal attributes with one-hot-encoding.

You can use Category Encoders package for use with scikit-learn in Python.

Read this blog for more approaches for data encoding - Smarter Ways to Encode Categorical Data for Machine Learning.

3. Install Weka system on your computer

Sort all the property samples by the property prices and divide the samples equally into 5 categories/classes: Top value, High value, medium value, low value, bottom value.

Apply the KNN algorithm of Weka with K=5 to 10 to classify the property instances into 5 classes. Calculate the accuracy for each K values.

Apply RandomForest algorithm of Weka and report the performance.

You need to split the whole dataset into training (66% samples) and testing datasets (34% samples). Do the random splitting 10 times to calculate the average accuracy.

from sklearn.model_selection import train_test_split

xTrain, xTest, yTrain, yTest = train_test_split(x, y, test_size = 0.2, random_state = 0)

	K=5	K=6	K=7	K=8	K=9	K=10
KNN	Average accuracy	...
RandomForest	Average accuracy

Write report to discuss the performances of KNN and randomforest. You are encouraged to compare the performance of different missing value imputation methods or the categorical encoding methods.

Attachment:- Data Mining & Warehousing Assignment Files.rar