AppliedMachineLearningPublic

Question 1: In Laurence Maroney’s video, What is ML, he compares traditional programming with machine learning and argues that the main difference between the two is a reorientation of the rules, data and answers. According to Maroney, what is the difference between traditional programming and machine learning?

Machine learning is using/teaching computers to model and discover relationships between data so that in the future a computer can predict an answer based on the previously studied data/relationship within the data. The difference between traditional programming and machine learning is with traditional programming the flow of data model has the rules and data as which are then used to produce answers, whereas machine learning turns this process around and instead uses data and answers to work backwards to figure out the rules of / relationship between the data.

Question 2: With the first basic script that Maroney used to predict a value output from the model he estimated (he initially started with 10 that predicted ~31. Modify the predict function to produce the output for the value 7. Do this twice and provide both answers. Are they the same? Are they different? Why is this so?

The predicted answers for the value 7 using Maroney’s script are not the same. The first run-through produced the answer 22.00052 while the second produced 22.000023. This slight difference is due to the fact that machine learning is based on probabilities: neural networks deal with parameters instead of hard coded values. So in this case, the neural network thinks that there is a high probability that the answer is 22, but the computer is not sure that the answer is exactly 22.

Question 3: Using the script you produced to predict housing price, take the provided six houses and train a neural net model that estimates the relationship between them. Based on this model, which of the six homes present a good deal? Which one is the worst deal? Justify your answer.

I think that the best deal in terms of the houses would be the Hudgins house because based on the relationship that the neural network picked up on the model predicted a price for this house that was above what it sold for, meaning that its actual asking value of $97,000 is less than what the model predicts it should be worth based on the number of rooms (it was predicted to be priced at $235,200) and the square footage it has (it was predicted to be priced at $155,100).

the worst deal would be either the church house or the moon house. The model predicted that the church house would have a value of about $299,500 based on how many rooms it has (4) and a value of $353,600 based on its square footage but its actual price is listed as $399,000, meaning that this house is way over-priced. The same goes for the moon house; based on the number of rooms it has the model predicted its value would be $170,900, and based on its square footage the model predicted its value would be $174,700, but its actual price is $250,000, so it is also overpriced.

House Actual Price # of rooms predicted price Difference sqft predicted price Difference
Church $399,000 $299,561 + $99,439 $353,633 + $11,190
Hodgins $97,000 $235,279 - $138,278 $155,171 - $201,853
Mobjack $289,000 $299,561 - $10,560 $340,955 - $98,809
Moon $250,000 $170,997 + $79,003 $1.74,568 + $40,102
Newptcomfort $229,500 $235,279 - $6,278 $285,366 - $69,853
Matthews $347,500 $363,843 - $16,342 $302,514 - $129,265