Presentation is loading. Please wait.

Presentation is loading. Please wait.

What determines height? Genetics NutritionGender Heels Child’s Height Parent’s Height Not able to get data for all our variables! Linear Regression vs.

Similar presentations


Presentation on theme: "What determines height? Genetics NutritionGender Heels Child’s Height Parent’s Height Not able to get data for all our variables! Linear Regression vs."— Presentation transcript:

1 What determines height? Genetics NutritionGender Heels Child’s Height Parent’s Height Not able to get data for all our variables! Linear Regression vs. Decision Trees How to compare and contrast algorithms? Research Goal Brainstorm & Lit Review Get & Prepare Data AnalysisAnswer

2 Linear Models vs. Decision Trees Height variable relationships appears linear Decision Trees Would Not Appear to be as Helpful

3 Holdout Some Data to Test Algorithms All Data Subset: Holdout ~1/3 Training Data ~2/3 Build Models Decision Trees Regression Evaluate Models

4 R Code for Creating Subsets # # Create Subsets # random_ids <- order(runif(934)) heights_train <- heights[random_ids[1:600],] heights_test <- heights[random_ids[601:934], ] The function runif creates random numbers The train and test lines use the array nature of dataframes (and programming in general)

5 Regression: Full Set vs. Training Data Child Height = 16. 5 + Father’s Height * 0.39 + Mother’s Height * 0.29 and If a Male then add 5.21 inches. Full Set Child Height = 16. 5 + Father’s Height * 0.39 + Mother’s Height * 0.32 and If a Male then add 5.21 inches. Training Models are fairly close

6 Decision Tree: Training Data

7

8 Decision Trees: Full Set vs. Holdout Data Models a little bit different

9 Comparison of Models (Subset of Data)

10 Correlations to Actual Value Heights DatasetLinear RegressionDecision Tree Initial Randomization0.8180.775 2 nd 0.7820.746 3 rd 0.8130.763 4 th 0.8420.826 5 th 0.7940.766


Download ppt "What determines height? Genetics NutritionGender Heels Child’s Height Parent’s Height Not able to get data for all our variables! Linear Regression vs."

Similar presentations


Ads by Google