Project Details

Explore insights into various factors contributing to accidents with the "Accident Analysis Dashboard.
The dashboard, accessible here, provides a comprehensive view of patterns and trends related to accidents.

Process:

Below is a summary of the process followed in developing the dashboard:

1. Imported the raw data.
2. Cleaned and preprocessed the data to ensure accuracy and reliability.
3. Created visualizations using the processed data to represent key insights.
4. Developed additional measures and columns to support visualizations.

Visualizations:

The dashboard includes the following visualizations:

• Accidents by Hour of Day
• Accident Severity by Road Type
• Number of Casualties by Vehicle Type
• Accidents by Weather Conditions
• Number of Vehicles Involved in Accidents
• Number of Accidents by Day of the Week
• Number of Accidents by Light Conditions
• Number of Accidents by Speed Limit

Project Details

The project methodology involves following steps:

1. Loading the Raw Data:

   The initial step involves loading the raw data to kickstart the analysis process.

2. Data Cleaning in Python:

   Data cleaning is crucial for ensuring the accuracy and reliability of the analysis.

   • Unnecessary columns are dropped to streamline the dataset.
   • Duplicate rows are removed to avoid data redundancy.
   • Individual rows are cleaned, ensuring data integrity.
3. Interactive Dashboard Development:

   With the data cleaned and prepared, the next step involves developing an interactive dashboard to visualize the insights gained from the analysis.

Key Questions and Analysis:

The analysis focuses on answering several key questions:

• Total Restaurants and Total Cuisines Worldwide: Exploring the overall count of restaurants and cuisines globally.
• Countries with the Greatest Number of Restaurants on Zomato: Identifying countries with the highest enrollment of restaurants on Zomato.
• Cities in India with the Greatest Number of "Value for Money" Restaurants: Analyzing which cities in India offer the best value for money dining options.
• Top 10 Cuisines with the Highest Number of Votes in India: Highlighting the cuisines that receive the most votes in India.
• Countries with Restaurants Providing Online Delivery: Identifying countries where restaurants offer online delivery services.
• Locality with the Most Restaurants in Terms of Quantity: Determining the locality with the highest number of restaurants.
• Restaurants with Good Cuisine and an Average Rating: Showcasing restaurants that excel in both cuisine quality and average rating.

Project Details

The project methodology involves the following steps:

• Data Acquisition: Acquired yearly Green House Gases Emission data from EDGAR website and daily temperatures data from kaggle.


• Data Preparation: Combined the emissions data with temperature data, taking variance, mean, maximum temperatures into consideration.


• Data Analysis: Conducted in depth data analysis including descriptive statistics and distribution plots of temperature features and emissions. Also calculated the top 10 countries with high emissions of both CO2 and GHG and analyzed sector-wise emissions. Performed correlation analysis, regression analysis, and time-series analysis of temperatures and emissions.


• Data Visualization: Created various visualizations to better understand the data and the relationships between variables. Also created scatter plots, line plots, heatmaps, and bar charts to visualize the data and observed patterns and trends.

Project Details

The project methodology involves the following steps:

1. Data Collection:
   • Imported essential libraries (NumPy, Pandas, etc.).
   • Understood data structure and types by examining columns.


2. Data Cleaning:
   • Cleaned raw data with 48,895 rows and 16 columns.
   • Removed null values and unnecessary columns for effective exploratory data analysis.


3. Exploratory Data Analysis:
   • Employed visualization techniques (bar charts, heatmaps, scatter plots, pie charts, violin plots, and multiple bar charts).
   • Explored and analyzed various features to gain valuable insights.

Key Insights:

• Most preferred locations: Manhattan and Entire home/apartment.
• Price distribution: Majority of listings fall within the 0 to 1000 range.
• Average prices: Manhattan has the highest (196.87), while Bronx has the least (87.50).
• Popular category: Entire home/apartment with the highest average price (211.81).
• Reviews: Brooklyn and Entire home/apartment lead in the number of reviews.
• Availability: Staten Island offers rooms with greater availability for more days.
• Special findings: 1,294 listings available for 365 days.

Project Details

The project methodology involves the following steps:

1. Data Collection:
   • Imported necessary libraries like NumPy, Pandas, etc.
   • Understood data structure, column meanings, and data types.
2. Data Preprocessing:
   • Cleaned the dataset with no null or duplicated values.
3. Exploratory Data Analysis:
   • Utilized exploratory data analysis techniques to understand patterns in the data.
4. Feature Engineering and Preprocessing:
   • Removed outliers and multicollinearity in the data.
   • Scaled input features using MinMaxScaler.
5. Regression Models:
   • Fitted the data into various regression models.
   • Evaluated metrics on the test dataset and plotted feature importance.

Conclusions:

• Most bikes are rented during the 18th hour, followed by the 19th hour, with the least rentals occurring during the 4th hour.
• In the summer season, bike rentals peak, while winter sees the least activity.
• On working days (No Holiday), the highest number of bike rentals occur, with Thursday having the highest count.
• In a typical functioning day, the bike rental count is at its peak.
• The distribution between numerical features and the dependent feature (Rented Bike Count) is spread throughout the entire area, indicating no specific linear relationship.
• Tree-based models outperform linear models, as independent features show non-linear relationships with the dependent feature ('Rented Bike Count').
• The optimal models are XGBoost Regressor and Gradient Boosting Regressor, demonstrating the required score without overfitting or underfitting.

Project Details

The project methodology involves the following steps:

1. Data Collection:
   • Imported necessary libraries like NumPy, Pandas, etc.
   • Understood data structure, column meanings, and data types.
2. Data Preprocessing:
   • Cleaned the dataset by handling null values, replacing with median and mode for numerical and categorical features, respectively.
3. Exploratory Data Analysis:
   • Applied exploratory data analysis techniques to gain insights into the data.
4. Feature Engineering and Preprocessing:
   • Removed outliers and multicollinearity from the data.
   • Scaled input features using StandardScaler.
5. Classification Models:
   • Fitted the training dataset into various classification models.
   • Evaluated models using the test dataset and plotted feature importance.

Conclusions:

By fitting data into various classification models and evaluating with the test dataset, the following conclusions were drawn:

• Recall and Precision:
  • For Case 1 (person has a disease but the model shows no risk), recall is the best evaluation metric.
  • For Case 2 (person has no risk of disease but the model predicts risk), precision is the best evaluation metric.
  • F1_score is suitable when both cases are important.
• Model Preferences:
  • If recall is important (focusing on Case 1), Logistic Regression and KNN models are preferable.
  • If precision is important (focusing on Case 2), Naive Bayes model is preferable.
  • If F_Score is important (focusing on both cases), Logistic Regression model is preferable.
• Feature Importance:
  • For all classification models, the 'age' feature has higher importance.

Project Details

The project methodology involves the following steps:

1. Data Collection:
   • Imported essential libraries (NumPy, Pandas, etc.).
   • Understood data structure, column meanings, and types.
2. Data Preprocessing:
   • Cleaned play store app data (10,841 rows, 13 columns) and user reviews data (37,427 rows, 6 columns).
   • Handled null values and duplicates, creating new features for analysis.
3. Exploratory Data Analysis:
   • Utilized visualization techniques (bar charts, heatmaps, scatter plots, distribution plots, and pie charts).
   • Extracted insights from various features to inform app-making strategies.

Key Insights:

• Categories:
  • Most apps fall under FAMILY, followed by GAME, with BEAUTY being the least popular.
  • EVENTS category boasts the highest ratings, while COMMUNICATION dominates in installs, and GAME in reviews.
• Updates and Pricing:
  • Majority of apps (66.97%) were updated in 2018, correlating with higher ratings, installs, and reviews.
  • 92.63% of apps are free, with 99.41% of installs coming from free apps.
• Ratings and Revenue:
  • Most ratings fall in the 4 to 5 range, and the majority of paid apps are priced between 0 to 50 USD.
  • LIFESTYLE category leads in average revenue, with few apps having significant earnings.
• Top Apps:
  • Top 5 revenue apps include Minecraft, I Am Rich, I Am Rich Premium, Hitman Sniper, and Grand Theft Auto: San Andreas.
  • Tools is the genre with the highest number of apps.
  • Facebook has the highest reviews, with 64.04% positive, 21.29% negative, and 14.67% neutral reviews.
  • The most popular app, based on positive reviews, is ColorNote Notepad Notes.

Project Details

Data Analysis Process:

The analysis involves the following key steps:

1. Imported and organized the raw e-commerce data in Excel.
2. Utilized Excel functions and tools for data cleaning and preprocessing.
3. Designed an interactive dashboard to visualize key insights.

Dashboard Components:

The dashboard features various components for a comprehensive analysis:

• Sales Analysis During Time: Visualizes sales trends over time.
• Delivery Analysis using Donut Chart: Illustrates delivery performance with a donut chart.
• Customer Acquisition using Waterfall Chart: Analyzes customer acquisition through a waterfall chart.
• Customer Satisfaction using Bar Chart: Represents customer satisfaction levels with a bar chart.

Key Insights:

Some key insights derived from the analysis include:

• Understanding sales patterns during different time periods.
• Evaluating delivery performance with a focus on timely deliveries.
• Analyzing the effectiveness of various customer acquisition channels.
• Measuring and improving customer satisfaction levels.