Download (right-click, save target as ...) this page as a Jupyterlab notebook from: ES-4


CE 5319 Machine Learning for Civil Engineers
Fall 2022 Exercise Set 4

LAST NAME, FIRST NAME

R00000000


Purpose :

Obtain databases for use in class and homework exercises.

Assessment Criteria :

Completion, results plausible, format correct, calculations (Jupyter Notebook) are shown.



Problem 1

Use the National Bridge Inventory Database and perform a rudimentary data analysis and content summary (just use ES-2 results).

Extract a Texas-subset, Extract a North Dakota-subset.

Use the North Dakota subset and make a classification model (logistic) that identifies failed (code 3 or smaller) and adequate bridges using the prediction variables described in Prediction of Bridge Component Ratings Using Ordinal Logistic Regression Model

Repeat with the Texas subset.

Comment on the performance of your model(s).


Problem 2

Use the Algerian forest fire dataset https://archive.ics.uci.edu/ml/datasets/Algerian+Forest+Fires+Dataset++

Make a classification model (logistic) that predicts fire/not-fire based on remaining variables.

Does region (one of the input variables) matter?

Suppose you make a model based on only a single region. How well does that model perform on the other region.?


Problem 3

Repeat the above problems (1 and 2) using KNN as the classification engine. Be sure to use the same training sets. Does KNN perform any better/worse (your opinion) than logistic regression?

In [ ]: