Normal Distribution of AQI Data

• Check the Continuous distribution of AQI dataset across the column.

• Applying Statistics & Visualization

• The AQI data distribution is approximately normal, indicating that the West Phoenix site has notably worse air quality than others, necessitating further examination and increased resource allocation for improvement.

Skills: Empirical Rule, Z-Score & Hist Plot

Python | Pandas | SciPy | Matplotlib

Built a Confidence Interval for AQI

• Constructing a Confident Interval of AQI dataset for RRE Operation

• Applying Sampling, EDA & Statistical Test

• The 95% confidence interval from the sample data yielded [10.36, 13.88], suggesting a notably greater population mean AQI for California with 95% certainty than 10.

Skills: Sampling, MOE, P-value & Box Plot

Python | Pandas | SciPy | Seaborn

Multiple Regression Estimation of sales

• Identifying the linear relationship between Sales from ads and Streaming Services Budget involves considering both continuous and categorical variables. 

• Achieved through Ordinary Least Squares (OLS) regression to model the data and validate the multiple regression assumptions using Visualization Plots and the VIF formula. 

• Using TV and Radio as the independent variables results in a multiple linear regression model with R²=0.904. In other words, the model explains 90.4% of the variation in Sales. Suggesting a strong relationship between the advertising expenses on TV and Radio and the Sales generated. 

Skills: OLS, EDA, Residuals, Q-Q Plot, scatter Plot, VIF.

Python | Pandas | SciPy | Seaborn | Matplotlib

Logistic Regression of Airways Data

• Determine statistical significance of Aircraft customer satisfaction with Inflight Entertainment, using satisfaction as dependent categorical variable and Inflight Entertainment as independent variable.  

• I split the dataset into test and train data, trained a logistic classifier model, and evaluated its performance using a Confusion Matrix to calculate metrics like Accuracy, Recall, Precision, and F1 Score. 

• Customers who rated in-flight entertainment highly were more likely to be satisfied, and logistic model achieved an accuracy of 80.2%, showing an improvement over the dataset's customer satisfaction rate of 54.7%. Interpretation of the Confusion Matrix shed light on false positives and false negatives.

Skills: Logistic Regression, Confusion Matrix, Recall, Accuracy.

Python | Pandas | Scikit learn | SciPy | Seaborn | Matplotlib

K-Means Clustering of FIFA Players 

• Developed a Python-based solution for clustering FIFA 22 players using K-Means. I loaded player data, implemented K-Means from scratch, and provided an alternative using Scikit-Learn. 

• Authenticated Google Colab, loaded player data from Google BigQuery, and I implemented K-Means manually with NumPy and visualized the process with PCA. Additionally, provided a Scikit-Learn alternative and visualized cluster  with inertia and silhouette scores. 

• My project enables users to cluster FIFA players based on attributes for insights into performance and potential. It offers flexibility with clustering methods and helps users make data-driven decisions in the game. 

Skills: SQL, K-Means, PCA, Silhouette score.

Python | Pandas | Scikit learn | BigQuery  | Seaborn

Movie Recommendation system 

• Developed a Python-based Content based movie recommendation system using TF-IDF and cosine similarity. Users input movie titles and receive real-time recommendations based on movie similarity. 

• I loaded movie data and implemented a search engine that cleans movie titles and calculates similarity. Interactive widgets allow user input, and a recommendation function finds similar movies based on user preferences. 

• Project enhances the movie-watching experience by providing personalized movie recommendations. The user-friendly interface fosters engagement. It showcases the application of NLP techniques for data exploration. 

Skills: TfidfVectorizer, widgets, ipython

Python | Pandas | Scikit learn | Numpy