Week 01 TODOs

[sayantikabanik]

Sayantika

• Basic EDA
• Read the articles
• Jot down my understanding

[sahithi02]

Gist from the recording of Abhay :

Three different aspects dimensions burnout

• Exhaustion:Mental exhaustion, overwork, responsibilities at home ,work, no bandwidth left

• Depersonalisation:You look at yourself as robot, detach yourself from the work

• Lack of achievement:Your level of achievement is less

• Our goal is to find many different ways to find where people are on this spectrum(like may be you are very exhausted but other facets are fine)

• How they combine make different profiles .

• Our goal is to assess where people are by measuring these aspects.

• Early efforts:Measurement and quantification.

1. Textual data analytics:Phase1

• Start with reviews corpus .Develop on the reviews ,eventually work on the peak health and mapping them to the data we have
• 3 different aspects have different facets such as Work life balance facet maps with exhaustion.
• 7 subcategories are like the facets in the shared data

Based on the underlying foundation we should be able to tell the state of workforce to the manager Most powerful state to summarise this data for the managers to understand.

2.Phase2:

• Deliverable1:Digital Watercooler- Harnessing the positives people are talking .Pros summarised effectively to the point where we can see them.How people have managed working well should be harnessed to show the employees feeling burned out End consumer:Employee
• Deliverable2: Suggestion box:Harnessing the cons ,dislikes into a digital suggestion box End consumer:HR

3.)Slice and dice: Manager of the group wants to know where the problems are happening, example same job function form different locations or different combinations of the Data.More like a comparator One for the employee and one for the HR Employee-Strength finder:Do I have others who are like me Heatmap kind of thing how others are feeling ad reporting the same and relevant to the employee

Hard core machine learning:Machine learning model which takes a review and predict either the overall score or the category We have the labelled data.

[AnizD]

Going through the HBR article. Will jot my points here.

[anuraagbhavaraju]

I started EDA on the CSV files. I shall post my analysis, relevant documents, python notebooks

[AnizD]

The HBR article mostly talks about the signs of burnout, the stressors in a job leading to burnout, how one can combat/prevent burnout and how leaders(and organizations) can help to solve this issue by putting sustainable measures in place.

Only a small(last) chapter is dedicated on measurement of burnout.

The suggestions highlighted can be a good consideration when we are building the suggestion box.

I have written this summary chapter-wise so that it will be easy to go back for any references.

Google drive link: https://docs.google.com/document/d/1CAzxbs6sSyOU1RZSSkbwkPIXyOVcua3B/edit?usp=sharing&ouid=100584766015578644570&rtpof=true&sd=true

[AnizD]

I will look into other articles around how burnout is measured and dig further into Maslach Burnout Inventory (MBI)

[sahithi02]

NasaZataDinaNyomanJuniartaJISEBI61.pdf

[anuraagbhavaraju]

Update on my progress so far:

• Created a Jupyter Notebook reviewsCombined.ipynb to combine all given datasets (HCl, TCS, Wipro, Infosys) into a single csv file

• Started EDA on entire data. Updated Data Cleaning.ipynb

Below are few key observations and relevant progress:

1. 30%-50% data missing in two important columns reviewText and location

2. Location (and even employee experience) is available in the jobFunction column. Extracted this information and populated required columns accordingly

Next Steps: Need to work on figuring out below items:

1. A lot of cleaning and harmonization is required for Job Function details
2. More than 50% reviewText data is missing which is key data for textual analysis