BUSI 651 – Individual Assignment 1
Machine Learning Algorithms
Note: Do not include the questions as well as dataset in your submission (to avoid similarity with
other submissions)
Question 1:
In this problem, we aim to develop a predictive model to estimate the energy consumption of
buildings equipped with HVAC (Heating, Ventilation, and Air Conditioning) systems. The
Dataset.csv contains information on various features that influence energy consumption in
buildings. The goal is to leverage these features to accurately predict the energy usage (in
kilowatt-hours) for different buildings. The table contains the following columns:
Room Area The area of the rooms in the building.
(sq. ft.)
Number of The count of appliances present in each building.
Appliances
Outside The outside temperature recorded at each building's location.
Temperature
(°C)
Insulation The thickness of insulation in the building's walls.
Thickness
(inches)
Building A categorical feature representing the type of building, such as residential or
Type commercial.
HVAC Another categorical feature representing the type of HVAC system installed in the
System building, such as Central AC, Split AC, or Window AC.
Average An additional numerical column highly correlated with the Outside Temperature,
Temperature reflecting a similar trend in terms of impact on energy consumption.
in last 24
hours (°C)
Energy The output variable, representing the actual energy consumption (in kilowatt-hours)
Consumption of each building based on the given features.
(kWh)
The objective of this problem is to build a predictive model that accurately estimates the
energy consumption of buildings with HVAC systems. By leveraging the provided features, the
model will predict the energy usage for new, unseen buildings, assisting homeowners,
businesses, and energy providers in optimizing energy consumption and reducing costs.
Required Analysis:
a) Determine the primary feature influencing energy consumption prediction? What about the
secondary feature? Explain the reasons behind their significance.
b) Identify any feature that may not contribute significantly to prediction accuracy. What is
your mitigation strategy? If no such feature found, provide justification for your claim.
c) Apply multiple linear regression to build a prediction model for energy consumption (y)
based on the features. Feel free to modify the dataset to enhance prediction accuracy. Use
the model to predict energy consumption for point1, point2, and point3:
point 1: Room Area (sq. ft.): 279, Number of Appliances: 16, Outside Temperature (°C): 20,
Insulation Thickness (inches): 1.7, Building Type: Residential, HVAC System: Central AC,
Average Temperature in last 24 hours (°C): 19, Energy Consumption (kWh): 385
point 2: Room Area (sq. ft.): 277, Number of Appliances: 22, Outside Temperature (°C): 15,
Insulation Thickness (inches): 1.5, Building Type: Commercial, HVAC System: Split AC,
Average Temperature in last 24 hours (°C): 14, Energy Consumption (kWh): 425
point 3: Room Area (sq. ft.): 276, Number of Appliances: 14, Outside Temperature (°C): 25,
Insulation Thickness (inches): 2.2, Building Type: Residential, HVAC System: Window AC,
Average Temperature in last 24 hours (°C): 26, Energy Consumption (kWh): 350
d) Compute the Mean Squared Error (MSE) regression loss using the "mean_squared_error"
function from “sklearn.metrics” library, for these points:
where y_true is ground truth (correct) target values and y_pred is estimated target values
based on the linear regression model. (Note that y_true and y_pred are vectors each
containing three values corresponding to point1, point2, point3.)
e) Given the following scatter plot of a new feature X with respect to Energy Consumption,
recommend whether to include it in the features for prediction. Provide reason.
600
550
Energy Consumption
500
450
400
350
300
250
40 45 50 55 60 65 70 75 80
New Feature
Question 2:
Consider the figure below where data points are divided into two classes, yellow circle, and
blue square. Please answer the following questions:
a) Draw the decision boundary for KNN-algorithm when K=1.
b) How will the following points be classified by K=3 classifiers:
Prediction
Blue Square
point (8, 6)
Yellow Circle
Blue Square
point (8, 4)
Yellow Circle
Question 3:
Given the following table where data points belong to two classes - and +. Using the KNN
algorithm to predict the label (+, -) for the following points:
- point1: (7.81, 5.33)
- point2: (9.43, 5.29)
a) What is the predicted label (+ or -) based on K=3?
b) What is the index of the closest neighbors for each point?
Index X1 X2 Y
1 8.27 5.59 +
2 1.58 5.87 -
3 5.92 5.87 -
4 9.44 5.83 +
5 2.11 5.57 -
6 4.71 5.94 +
7 3.82 5.84 +
8 6.98 5.91 -
9 3.15 5.42 -
10 8.9 5.94 -
11 7.65 5.77 +
12 9.83 5.29 -
13 1.94 5.36 +
14 7.13 5.28 -
15 5.77 5.47 -
16 4.36 5.31 +
17 5.09 5.65 -
18 3.42 5.24 +
19 2.76 5.71 +
20 9.6 5.52 -
Delivery:
For the final submission, create tables like this and fill them with your results/answers:
Question 1:
a) primary feature: secondary feature:
reason:
b) feature not contributing:
mitigation strategy:
If no such feature exists, justify your claim:
c) Apply multiple linear regression:
energy consumption for point 1:
energy consumption for point 2:
energy consumption for point 3:
d) Mean Squared Error (MSE) regression loss:
e) recommendation to include the new feature (with reason):
Question 2:
Decision Boundary for K=1:
Classify:
Prediction
Blue Square
point (8, 6)
Yellow Circle
Blue Square
point (8, 4)
Yellow Circle
Question 3:
Predicted labels (+ or -):
Point 1:
Point 2:
Index of closest neighbors:
For Point 1:
For Point 2:
1.0 Grading Rubric
Key Points Grade Allocation
(%)
Format (font type, size, table, formulas), overall 10
content, including references if required (APA
Style)
Results, analysis and assumptions 80
Novelty and creativity in solution 10
N.B. Failure to comply with the above would result in low grades.
2.0 Format and Deadline
Submission: a single word/pdf document
Deadline: As posted in Course Shell
