Climate Change Group Project

Team

• Irene Depacina
• Sebastian Villanoy
• Stephanie Juniper
• Vinutha Phani

Selected Topic : "Climate change by State in the USA"

Rationale

• Climate change is an interesting topic with lot of available data.
• US is one of the largest emitters of CO2 in developed countries.
• Population density and land mass lends itself to analysis
• Dataset availability

Discovery Questions

• Can we quantify the human impact on climate change in the USA?
• Does human impact affect how states individually experience climate change?
• Are there climate change trends evident in temperature over time?
• Are some states exacerbating climate change through consumption patterns?
• Using the datasets, can we predict temperature using human impacts on climate change versus geographical information alone?
• Is there a possibility to predict any other feature like CO2, year, state disasters from the datasets?

Key Findings from the analysis

• A model that solely uses human-related features to predict climate change measured in average annual state temperature is not accurate
• Location has major influence both temperature and human-related features
• A model that quantifies the impact of a state's energy consumption pattern on CO2 emissions, the main culprit of rising temperatures
• Energy consumption from non-renewable sources such as coal, petroleum, and natural gas contribute to increasing CO2 emissions
• Energy consumption from renewable sources such as biomass, hydropower, solar, and wind either reduce CO2 emissions
• While humna impact on temperature is difficult to determine concluively with our data, human impact on CO2 emissions can be predicted and impacted through responsible selection of energy sources.

Dashboard

• Visualizations were created using Tableau and HTML Hybrid.
• HTML/CSS/Java webpage Climate Change Webpage
• Interactive Tableau Visualizations
  • Climate Change Trends over Time
  • Results by State
  • Unsupervised Machine Learning Model Clusters
  • Machine Learning Predictions
• Please find presentation of our analysis on Google Slides
• Please see Dashboard --> README for more details.

Data Sources:

• While in search of climate change related datasets for US, ended up identifying some of the sites to have reliable datasets in csv format.
• Saved these datasets into repository under Resources/data_raw.
• These datasets are from websites :
  • https://data.world/environmentdata/
  • https://www.eia.gov/
  • https://www.ncdc.noaa.gov/
• Please see Data_Processing --> README for more details.

Database

• Datasets of .csv file format, hence decided on Postgres DB, using SQL and PgAdmin
• Total of 10 tables were designed, created and loaded into climate_change PostgreSQL DB.
• Please see Database --> README for more details.

Machine Learning

• Unsupervised Machine Learning performed as exploratory machine learning using PCA, Elbow curve & KMeans Algorithm
• Multiple linear regression using Neural Network and SKLearn
• Multiple linear regression models using R to identify confounding features and features behaving as proxies for US states
• Detailed statistical analysis using various data transformation techniques
• Please see Machine_Learning --> README for more details.

---

---