Scribd
Upload
3 views5 pages

Assignment - Module 2

The assignment involved analyzing customer reviews as a customer service analyst, comparing manual sentiment ratings with those generated by an AI model. Initially, the AI misclassified all reviews as neutral, but after refining the prompt, its performance improved significantly. The results were mostly aligned, though the AI struggled with detecting neutral sentiments, highlighting the importance of prompt design and human input in AI analysis.

Uploaded by

babdulrahman822
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
3 views5 pages

Assignment - Module 2

The assignment involved analyzing customer reviews as a customer service analyst, comparing manual sentiment ratings with those generated by an AI model. Initially, the AI misclassified all reviews as neutral, but after refining the prompt, its performance improved significantly. The results were mostly aligned, though the AI struggled with detecting neutral sentiments, highlighting the importance of prompt design and human input in AI analysis.

Uploaded by

babdulrahman822
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

‭Assignment - Module 2‬

‭Applying Machine Learning in Real-World Data‬


‭By Bello Abdulrahman:‬

‭ erhaps this is the most exciting assignment so far. After completing Module 2, which focuses‬
P
‭on machine learning, I started the practical assignment. The task required me to take on the role‬
‭of a customer service analyst. I manually analysed customer reviews and assigned sentiments‬
‭of positive, negative, or neutral to each one. I then compared my analysis with the results‬
‭generated by an AI model to evaluate its performance.‬

‭ or the first attempt, I manually created an Excel file with data from 10 customers, including‬
F
‭their names, cities, and reviews of a pair of headphones. I saved this as my original file. Then, I‬
‭made a copy and added my own sentiment ratings to the second [Link] to the second‬
‭version.‬

‭Here are my added sentiments‬


‭ ext, I asked the ChatGPT to analyze the reviews and determine the sentiment. However, the‬
N
‭initial result was unsuccessful, the AI classified all reviews as “Neutral”, which was clearly‬
‭inaccurate.‬

‭ o improve the outcome, I revised the prompt I gave to ChatGPT. This time, the results were‬
T
‭more plausible and accurate, showing that prompt design plays an important role in AI‬
‭performance.‬
‭ hen comparing my manually analyzed data with the AI-analyzed data, I noticed that our‬
W
‭results were almost entirely similar, with only a few differences. Specifically, in three instances,‬
‭the AI was unable to correctly identify a neutral sentiment — instead, it labeled those‬
‭statements as positive. This shows that while the AI performed well overall, it struggled slightly‬
‭with detecting neutral tones in certain reviews.‬
‭ he AI mostly matched my labeling. Apart from a few cases where it misread neutral statements‬
T
‭as positive, the rest of the results were pretty much the same as mine. That small difference‬
‭showed me that while the AI is good, it still misses some of the subtle tones in customer‬
‭reviews.‬

‭ ne challenge I faced was getting the AI to respond properly at first. My initial prompt didn’t‬
O
‭work well—it labeled everything as neutral. But after tweaking the prompt, the results got way‬
‭better. This activity really helped me understand how important the prompt is when working with‬
‭AI, and it also showed me that human input is still very important when analyzing things like‬
‭customer feedback.‬

You might also like