odeyiany2 / ecx-4.0-21-days-data-science-challenge

ECX 4.0 21 Days Data Science Challenge: Iris Classification

This is a challenge organized by the Engineering Career Expo Unilag. A challenge to test our data science skills for 21 days. In this challenge, I dealt with the Iris dataset and built a model that predicts the specie of an Iris by its measurements.

Tools

The following tools were used for different areas of the project:

• Python Libraries:

  • Pandas: for data analysis and manipulation
  • Seaborn: a library based on matplotlib and it provides a high-level interface for data visualization
  • matplotlib: for data visualization
  • Joblib: Saving our model for deployment

• Scikit Learn (Python Machine Learning Library):

  • GridSearchCV and RandomSearchCV: Hyperparameter tuning
  • StandardScaler: for standardization of numeric features
  • LabelEncoder: for encoding oyr categorical features
  • RandomForestClassifier, SVC, LogisticRegression, DecisionTreeClassifier: ML algorithm for classification problems

• Evaluation Metrics:

  • Accuracy Score: Number of correctly predicted class over the total classes
  • Precision: ratio of correctly predicted positive classes over the total positive classes
  • Recall: ratio of correctly predicted positive class over the total classes
  • Classification report: a report showing precision, recall and F-1 score
  • ROC Curve: a plot showing the true positive rate(TPR) over false positive rate(FPR)
  • Confusion matrix: a table for assessing the quality of our classification model prediction

• Deployment: Streamlit

Deployment Video

Streamlit App