The automobile industry is flooded with options, it is a dream for a middle-class family to buy a car, and with this incentive, the company is doing everything possible to give a smooth ride utilising analytics. As a result, I'm attempting to develop an analytical tool to provide Data-Analysis for User given Dataset for the automotive sector as per Manufacturing Industry Employees as a user to take informed decisions.
As the manufacturing industry must make selections based on car reviews. The dataset is being harmed by FAKE/Misleading Reviews, which is a big problem. It must be identified and extracted from the data in order to make excellent business decisions. Hence my Project will also solve this Problem by doing Real-time Fake Review Detection.
- Industry Employees are the User for the Project to Take informed Decisions by this tool.
https://www.youtube.com/watch?v=JZ0lHTFIjoo
- Django
- Html/css/javascript
- Dbsqlite Database
- bootstrap
- jupyter-notebook(webscraping using beautiful-soop)
- seaborn for Visualization
Install Requirnment.txt file using Pip. Run pip install -r requirements.txt
-
Clone the repo
-
Cd (check where manage.py must be present)
-
Use python manage.py makemigrations
-
Followed by python manage.py migrate
-
The project setup is completed and ready to start. Use python manage.py runserver to Start the project in local Host.
-
Home Page - DashBoard
-
Fake-Review Detection (Real-time)
- Web Scraping From Amazon review to Train model.
-
Data Analysis Tool (for custom dataset)
- Exploratory Data-Analysis
- Cluster Analysis
- Correlation Analysis
-
Command-line Query for Generating graphs
-
SignUp/SignIn (for particular user)
Fake reviews make it extremely difficult for manufacturers to make informed judgments, therefore I decided to write a function to detect and remove fake reviews from the dataset for accurate demand and feature forecasts.
-
TEXT box where user can Write its Query Whether it is Fake or Not / also can insert Fake Review excel dataset
-
Pre-processing of Text as per user.
-
Inline Character as well as Word Count feature
-
-
Result Page with decisions:
-
In addition, I will provide a default analysis of the given dataset, including client groups, the most popular automobile specification combinations (engine type, fuel, mileage, and so on), the ideal time to introduce a new car, and so on. as it is capable of:
-
After that, the user must Insert Dataset. It will take the user to the next page, where they can view the dataset and its features.
-
Three options are available in the navigation bar. This will traverse according to the user's actions
- Histogram of Price
- Dominating car BodyType
- BoxPlot for Price (Outlier analysis)
- engine size comparision
- Relationship for Price and Power
- Cluster the cars types and cars using k-means algorithm
- Price and horse power with cluster price
- Power and Mileage after clustering
- Engine size with Fuel tanks
- Average price with each cluster
- Finding potential stretegic groups
- Cars body type with each cluster
- Correlation Matrix (to know which features all strongly correlated)
- Extensive scatter plot grid of more numerical variable to investigate the realtion in more detail
- A command line tool where the user can dynamically enter a query.
- the required visualisation will be generated with its inference, allowing the user to understand what the graph is doing.
- Currently which option is limited to only 3 graphs as per prototype(line,scatter,bar)
