Ikuku-Success / Udacity-Data-Analytics-Nanodegree-Project-II

The dataset is the tweet archive of Twitter user @dog_rates, also known as WeRateDogs. WeRateDogs is a Twitter account that rates people's dogs with a humorous comment about the dog. These ratings almost always have a denominator of 10. The numerators, though? Almost always greater than 10. 11/10, 12/10, 13/10, etc.

ALX-Udacity Data Analytics Nanodegree Project-II

📝Introduction

The dataset is the tweet archive of Twitter user @dog_rates, also known as WeRateDogs. WeRateDogs is a Twitter account that rates people's dogs with a humorous comment about the dog. These ratings almost always have a denominator of 10. The numerators, though? Almost always greater than 10. 11/10, 12/10, 13/10, etc. Why? Because "they're good dogs Brent." WeRateDogs has over 4 million followers and has received international media coverage.

📝Data Cleaning & Transformation

Quality issues

• There are columns with missing values namely - expanded_urls

• Name, doggo, floofer, pupper and puppo columns have value with the name None

• Values in source column are not human readable

• Use of _instead of space in p1, p2 and p3 column values. Also, values upper case sometimes lowercase other times in p1, p2 and p3 columns

• column names are not clearly descriptive

• There are image duplicate predictions present for duplicate jpg_url with different tweet ids and rest all the data same.

• tweet id title is different, id here tweet_id in others.

• There are image duplicate predictions present for duplicate jpg_url with different tweet ids and rest all the data same

Tidiness issues

• One variable in 4 columns - dog_stage(doggo, floofer, pupper, puppo)

• Merge the tables - archive_tweet and df

📝Insights

• Comparison Of Average Number Of Retweets And Favorites In A Month
• Comparison Of Average Rating Of Dog Stages
• Algorithm's First Prediction Efficiency for top 10 most frequent predictions

📝Conclusion & Observation

• Algorithm has proved to be least efficient for Chihuahuas.
• The most predictions are recorded for Golden Retrievers and least for Malamutes
• Therefore, we can say that, this algorithm favors the Golden Retrievers in this dataset
• From the bar graph, it can be clearly seen that puppo has the highest average rating while pupper has the lowest.
• From the pie chart, we can see that there are 63.48% of puppers in the overall population of the dataset followed by doggo, then floofer and at the last, puppo. - - This indicates the possible reason of lowest average rating of pupper and highest average rating of doggo because of huge differences in their proportion in the dataset
• As we can see that on an average, number of favorites are quite higher in January.
• February has the second highest both average number of favorites and retweets.
• December has the least number of favourites which is in close range with september, october & november.