We will use one full day worth of tweets as our input $($there are total of $4\mathrm{.}4$M tweets in this file$)$:

Execute and time the following tasks with $110,000$ tweets and $550,000$ tweets:

a$.$ Write and execute a SQL query to find the average longitude and latitude value for each user ID$.$ This query does not need the User table because User ID is a foreign key in the Tweet table. E$.$g$.,$ something like SELECT UserID, MIN$($longitude$),$ MAX$($latitude$)$ FROM Tweet, Geo WHERE Tweet.GeoFK $=$ Geo.GeoID GROUP BY UserID;

b$.$ Re$-$execute the SQL query in part $2-$a $5$ times and $20$ times and measure the total runtime $($just re$-$run the same exact query multiple times using a for$-$loop, it is as simple as it looks$).$ Does the runtime scale linearly? $($i$.$e$.,$ does it take $5$X and $20$X as much time?$)$ What is the average runtime of each individual run?

c$.$ Write the equivalent of the $2-$a query in python $($without using SQL$)$ by reading it from the file with $550,000$ tweets.

d$.$ Re$-$execute the query in part $2-$c $5$ times and $20$ times and measure the total runtime. Does the runtime scale linearly? What is the average runtime of each individual run?

e$.$ Write the equivalent of the $2-$a query in python by using regular expressions instead of json.loads$().$ Do not use json.loads$()$ here. Note that you only need to find userid and geo location $($if any$)$ for each tweet, you don$$t need to parse the whole thing.

f$.$ Re$-$execute the query in part $2-$e $5$ times and $20$ times and measure the total runtime. Does the runtime scale linearly?

g$.$ Create a visual using matplotlib of $2$d showing the distribution of the runtimes.