This Side of Paradise

An analysis of architects' salary while I'm preparing for interview

🔨 Under Construction Now 🏗️

I make this repo public to be accountable, hence it will be a hot mess for one week or so :)

Repo name origin

"This Side of Paradise" is a novel by American writer F. Scott Fitzgerald, a major theme in which is the disillusionment one feels as one grows in his college years.

This theme properly describes a lots of architects' feelings when they find the disillusionment between their salary and once aspiration, me included, two years ago.

Aims

1. Find insights and patterns of architects' salary.
2. Revisit statistics knowledge in practice.
3. Illustrate why I'm leaving architecture/ urban planning industry.

Note

Definition of architects here: includes landscape architects, urban designers, and urban planners, etc. Because (at least in China) these majors are all under the architecture category in universities.

Data source

1. A shared excel file, architects inputting their salary-related information anonymously.

The original data is not MIT License and I don't own the right to commit it in this repo. If you are interested in this data, please directly contact WeChat Official Acount 建筑透明性 at the bottom of this blog.

2. cn_stopwords.txt (Chinese stopwords in the NLP part) - goto456.2020.stopwords.https://github.com/goto456/stopwords/

3. centroids.md (Chinese province centroids in the interactive map part) - siliushi.2015.geocoord.https://github.com/siliushi/geocoord

Techniques used

data wrangling: Python - Pandas, Re(Regular Expression)

EDA: R, Python - Pandas

statistical model(multivariate regression): R

machine learning(NLP with TF-IDF and K-means): Python - scikit-learn

data visualization: Python - Folium, Matplotlib, Wordcloud

Estimate output

• final output

1. multivariate regression analysis, answering the question "what factor most influences architects' salary?"

2. cluster analysis of comments, answering the question "what are major salary-related topics architects talking about?"

3. good old web map, answering the question "how do architects' location distributed?"

• process output

1. structured salary data(not committed)