Description

Air pollution is a persistent problem of the world, which has been around for a while and will continue to exist in the future. It is a cause of a lot of problems and has extremely adverse effects. For curbing air pollution, its analysis is first necessary. For making predictions about the air quality, domain knowledge is essential, as are data analysis skills. This problem statement tests them all.

Problem Statement
Participating teams are invited to propose and implement a multiclass classification technique for a given data set of air quality measurements. We expect a well-structured report detailing the approach used for classification and its implementation. Teams would be provided with the training data set to be used. Teams are expected to brainstorm, ideate, experiment, and code classification techniques to get the best results. The goal of this challenge is to create awareness about the applications of data analysis and machine learning in the chemical engineering domain, especially air quality analysis for air pollution control applications among the student community and provide them a platform to showcase their ideas and innovations.

Data Set Information

Attribute Information
2. Time (HH.MM.SS)
3. True hourly averaged concentration CO in mg/m^3 (reference analyzer)
4. PT08.S1 (tin oxide) hourly averaged sensor response (nominally CO targeted)
5. True hourly averaged overall Non Metanic HydroCarbons concentration in microg/m^3 (reference analyzer)
6. True hourly averaged Benzene concentration in microg/m^3 (reference analyzer)
7. PT08.S2 (titania) hourly averaged sensor response (nominally NMHC targeted)
8. True hourly averaged NOx concentration in ppb (reference analyzer)
9. PT08.S3 (tungsten oxide) hourly averaged sensor response (nominally NOx targeted)
10. True hourly averaged NO2 concentration in microg/m^3 (reference analyzer)
11. PT08.S4 (tungsten oxide) hourly averaged sensor response (nominally NO2 targeted)
12. PT08.S5 (indium oxide) hourly averaged sensor response (nominally O3 targeted)
13. Temperature in °C
14. Relative Humidity (%)
15. AH Absolute Humidity

Target Information
CO level is given as five broad categories Very High, High, Moderate, Low and Very Low. The target is to predict this class of CO level based on all the attributes listed

Timeline & Submission Details
● Training dataset is made available her e
● Report + Source code should be submitted to technicalaffairs@iitp.ac.in with subject

Submission Guidelines
● Participating teams are expected to use any of the following programming languages for implementation:
○ Python
○ Java
○ Matlab
○ C
○ C++
○ R
TensorFlow, Caffe, Theano, Keras, PyTorch, etc
● The source code should be appropriately commented and must be accompanied by a ‘Read-Me’ file containing instructions to run the code. The “Read-Me’ files must also specify any additional packages/resources if used. Please provide the link to download the same
● Participating Teams are expected to submit a single .zip file containing the following:
○ Source Code Files
○ Read Me File
○ Classified Output (.csv)
○ Report (.pdf)
● Naming Convention: The .zip file must have the same name as your <house>.zip.
● The report is expected to follow the given format: ○ Team Details
■ Names and Contact details of the participants
○ Introduction
○ Describe the problem statement and the need for air quality analysis
○ Classification Approach
○ Motivation
○ Methodology
○ Implementation
○ Results
■ F1-score
■ Confusion Matrix
■ Kappa Coefficient
■ Overall Accuracy
○ Conclusion

Judging Criteria

Rules and Regulations
● Max Team size: 4
● The submissions will be scrutinized for forgery. Any sort of ethical misconduct will not be tolerated and will result in the disqualification of the team
● In case of any dispute, the decision of the judges or the expert panel will be final and binding on all.
● The team must adhere to the spirit of healthy competition.