We use two datasets - plants-based, and bank-note authentication datasets for the following analysis methods.
Plants-based dataset: data.txt file from https://plants.sc.egov.usda.gov/java/stateDownload?statefips=US04
Bank-Note Authentication dataset: bill_authentication.csv from https://archive.ics.uci.edu/ml/datasets/banknote+authentication
Required files are included this repository.
Big Data Assignment - descriptive.ipynb ; Required file: data.txt
Here, the plants-based dataset is analysed and pre-processed with techniques like integer encoding, data imputation, feature selection and data visualisation methods. Processed dataset saved in plants_data.csv .
Big Data Assignment - predictive.ipynb ; Required file: plants_data.csv
Classification algorithms are used to predict the Family attribute of a plant. Techniques used include feature scaling, SVM classifier, DBSCAN clustering, hyperparameter tuning to optimise learning parameters, and itemset mining methods.
Big Data Assignment - Rules and Patterns.ipynb ; Required file: plants_data.csv
Using the processed plants dataset, Apriori algorithm is implemented to find frequent itemsets. Further, "interesting-ness" of the rules and petterns generated using various measures is validated.
Big Data Assignment - Decision Tree and Bayes Classifier.ipynb ; Required file: bill_authentication.csv
Decision Tree and Bayes Classifier are used to predict whether a bank note is genuine or forged.