1 of 16

Find your own cocktail!

MadData 2024

Madison

Haneul Kim, Sodam Na, Clara Kim, Yein Choi

02/25/2024

University of Wisconsin

PRESENTED BY

DATE

COCKTAIL MATCH

2 of 16

1

ABOUT US

Haneul Kim	Sodam Na
Yein Choi	Clara Kim

3 of 16

TABLE OF CONTENTS

ABOUT US

COSINE SIMILARITY

DATA INTRODUCTION

DATA CRAWLING & CLEANING

SIMULATION

VISUALIZATION & ANALYSIS

Q & A

1

5

2

6

3

7

4

4 of 16

SIMULATION

2

5 of 16

3

DATA INTRODUCTION

6 of 16

3

DATA INTRODUCTION

7 of 16

4

DATA CRAWLING

8 of 16

4

�DATA CLEANING / COSINE SIMILARITY

After cleaning the data, we processed our recommendation algorithm using cosine similarity.

We repeated through different columns and combinations to get the highest score, and extracted the desired matrix to vectorize using TFIDF Vectorization. We transformed and fit the data, cleaned up all values to ensure the code run smoothly.

As a result, we came up to a decision to regard [ Flavor, Color, Main spirit, Season, Occasions ] columns in our recommendation system.

TFIDF: Term Frequency — Inverse Document Frequency

measures the frequency & importance of a term to create a numerical representation of a document. Each document is represented as a vector where each element corresponds to the TF-IDF score of a term in the document. This representation captures both the local importance (how often a term appears in the document) and global importance (how unique the term is across all documents) of terms in a document.

- measures the cosine of the angle between two vectors, providing a measure of similarity irrespective of their magnitude.

9 of 16

Triggering user input

5

10 of 16

VISUALIZATION

Q. Top Main Spirit?

6

11 of 16

VISUALIZATION

Q. Top 5 Ingredients?

12 of 16

Q. Correlation of Color and Flavor?

13 of 16

Q. Correlation of Season and Flavor?

14 of 16

Q. Color Trend by Season?

15 of 16

Q & A

7

16 of 16

Thanks !