def plot_top_by_tweet_count(max_films=10, crtiical_period=True):
"""
Function to plot films with the top tweet counts
:param max_films: integer for the maximum films to plot
:param critical_period: bool indicating whether or not to only look at tweets in the critical period
"""
fig = plt.figure()
ax = fig.add_axes([0, 0, 1, 1])
#count tweets over defined time scale for each movie
column = "tweet_count"
if crtiical_period:
movies_df["critical_period_tweet_count"] = movies_df.apply(
lambda row: movie_helper.count_tweets(row["movieId"], row[
"critical_start"], row["critical_end"])['count'],
axis=1)
column = "critical_period_tweet_count"
else:
movies_df["tweet_count"] = movies_df["movieId"].apply(
lambda x: movie_helper.count_tweets(int(x))['count'])
#sort in descending order by tweet count and take the max films
sorted_df = movies_df.sort_values(by=column,
ascending=False).head(max_films)
#create bar plot
ax.bar(sorted_df["title"], sorted_df[column])
ax.set_ylabel("Tweet Count")
ax.set_title("Top {0} by tweet counts".format(max_films))
plt.xticks(rotation=90)
plt.show()
def plot_tweets_vs_finance(column,
title,
xlabel,
ylabel,
movie_run=False,
logx=False,
logy=False):
"""
Function to plot the tweet counts vs finance column for movies df
:param column: string column name of variable to plot
:param title: string title for plot
:param xlabel: string x label for plot
:param ylabel: string y label for plot
:param movie_run: bool indicating if tweets should only be counted over the cinema rub
:param logx: bool indicating wether to use log scale on x axis
:param logy: bool indicating wether to use log scale on y axis
"""
#normalize column into millions of usd
movies_df["temp_col"] = movies_df[column].replace(
'[\£,]', '', regex=True).astype(float) / 1000000
#check if we need to only look at the movie run
if movie_run:
#do a loop?
movies = movie_helper.gen_movies(movies_df)
tweet_counts = []
for movie in movies:
tweet_counts.append(movie.get_geotweet_count_by_dates())
movies_df["tweet_count"] = tweet_counts
else:
#count all tweets
movies_df["tweet_count"] = movies_df["movieId"].apply(
lambda x: movie_helper.count_tweets(int(x))['count'])
#create scatter plot with regression line
ax = sns.regplot(x="temp_col", y="tweet_count", data=movies_df)
#check axis log sclaes
if logx:
ax.set_xscale('log')
if logy:
ax.set_yscale('log')
ax.set(xlabel=xlabel, ylabel=ylabel)
plt.title(title)
plt.show()
def correlatte_movie_stats():
"""
Function to create correlation of base stats for movies
"""
#https://towardsdatascience.com/better-heatmaps-and-correlation-matrix-plots-in-python-41445d0f2bec
#get all needed tweet columns
movies_df = movie_helper.get_movies_df_with_opening_weekend()
movies_df["tweet_count"] = movies_df.apply(
lambda row: movie_helper.count_tweets(row.movieId)['count'], axis=1)
movies_df["budget_usd"] = movies_df["budget_usd"].replace(
'[\£,]', '', regex=True).astype(float) / 1000000
movies_df["uk_gross_usd"] = movies_df["uk_gross_usd"].replace(
'[\£,]', '', regex=True).astype(float) / 1000000
movies_df["domestic_gross_usd"] = movies_df["domestic_gross_usd"].replace(
'[\£,]', '', regex=True).astype(float) / 1000000
movies_df[
"worldwide_gross_usd"] = movies_df["worldwide_gross_usd"].replace(
'[\£,]', '', regex=True).astype(float) / 1000000
movies_df["international_gross_usd"] = movies_df[
"international_gross_usd"].replace('[\£,]', '',
regex=True).astype(float) / 1000000
movies_df["gross_profit_usd"] = movies_df["gross_profit_usd"].replace(
'[\£,]', '', regex=True).astype(float) / 1000000
movies_df["opening_weekend_takings"] = movies_df[
"opening_weekend_takings"].replace('[\£,]', '',
regex=True).astype(float) / 1000000
#list of columsn to use for correlation
columns = [
'budget_usd', 'uk_gross_usd', 'domestic_gross_usd',
'worldwide_gross_usd', 'international_gross_usd', 'gross_profit_usd',
'return_percentage', 'uk_percentage', 'tweet_count',
'total_release_weeks', 'first_run_weeks', 'best_rank',
'weekends_at_best_rank', 'weekends_in_top_3', 'weekends_in_top_5',
'weekends_in_top_10', 'weekends_in_top_15', 'opening_weekend_takings',
'run_up_tweets', 'opening_tweets'
]
#generate heatmap
generate_heatmap_from_df(df, columns)
def gen_bottom_20_tweet_count():
"""
Function to plot the bottom 20 movies by tweet count
"""
#get movies from db and count tweets
movies_df = movie_helper.get_movies_df()
movies_df["tweet_count"] = movies_df.apply(
lambda row: movie_helper.count_tweets(row['movieId'])['count'], axis=1)
#sort values and take bottom 20
movies_df = movies_df.sort_values(by='tweet_count').head(20)
#do bar plot
plt.barh(movies_df["title"], movies_df["tweet_count"], color='green')
plt.ylabel('Movie Title')
plt.xlabel('Tweet Count')
plt.title('Bottom 20 Movies')
plt.show()