CODEWITHSUNDEEP
pythonpandasindexingmapping
AdrianC
AdrianC

Reputation: 393

Value_Counts, Mapping, GroupBy and Plotting

I am working on a personal project using IMDB Data and have currently exhausted all avenues.

Quick Overview:

So far, I have executed the following:

plt.subplot2grid((2,3),(0,1))
actor_1 = df.pivot_table(index="Actor_1", values="Gross_Earnings", aggfunc='sum').sort_values(ascending=False)
actor_1[:15].sort_values(ascending=True).plot(kind='barh', width=0.7, alpha=0.5, color='red')
ax.tick_params(axis='x', labelsize=60)
ax.tick_params(axis='y', labelsize=60)
plt.xlabel("Gross Earnings")
plt.tight_layout()

plt.show()

This works, but it only returns the top values; not the top values with the additional criteria of > 4 Films.

I have also tried the following:

no_of_films = df.groupby("Actor_1")
name_count_key = df["Actor_1"].value_counts().to_dict()
no_of_films["Films"] = no_of_films["Actor_1"].map(name_count_key)

But it returns the following error: "AttributeError: Cannot access callable attribute 'map' of 'SeriesGroupBy' objects, try using the 'apply' method"

no_of_films = df.groupby("Actor_1")
name_count_key = df["Actor_1"].value_counts().to_dict()
no_of_films["Films"] = no_of_films["Actor_1"].apply(name_count_key)

But it returns the following error: "TypeError: unhashable type: 'dict'"

The group-by function idea was to create a new column called "Films" so count the volume of Films each actor has starred in and then use > 4 but it returns bools and not the actual value.

Director        Actor_1         IMDB_Score   Gross_Earnings    Movie_Title
Andrew Stanton  Daryl Sabara    6.6          73058679          John Carter
Sam Raimi       J.K. Simmons    6.2          336530303         Spider-Man 3
Nathan Greno    Brad Garrett    7.8          200807262         Tangled
Joss Whedon     Chris Hemsworth 7.5          458991599         Avengers: Age of Ultron

Is this possible or am I being silly?

Any help would be greatly appreciated.

Thanks,

Adrian

Upvotes: 0

Views: 474

Answers (1)

jezrael
jezrael

Reputation: 862511

I think you need filter or boolean indexing with transform:

print (df)
            Director          Actor_1  IMDB_Score  Gross_Earnings Movie_Title
0      James Cameron      CCH Pounder         7.9       760505847      Avatar
1      James Cameron      CCH Pounder         7.9       760505847     Avatar1
2      James Cameron      CCH Pounder         7.9       760505847     Avatar2
3      James Cameron      CCH Pounder         7.9       760505847     Avatar3
4     Gore Verbinski      Johnny Depp         7.1       309404152     Pirates
5         Sam Mendes  Christoph Waltz         6.8       200074175     Spectre
6     Gore Verbinski      Johnny Depp         7.1       309404152    Pirates1
7         Sam Mendes  Christoph Waltz         6.8       200074175    Spectre1
8  Christopher Nolan        Tom Hardy         8.5       448130642         The

df1 = df.groupby(["Actor_1"]).filter(lambda x: len(x) > 3)
print (df1)

        Director      Actor_1  IMDB_Score  Gross_Earnings Movie_Title
0  James Cameron  CCH Pounder         7.9       760505847      Avatar
1  James Cameron  CCH Pounder         7.9       760505847     Avatar1
2  James Cameron  CCH Pounder         7.9       760505847     Avatar2
3  James Cameron  CCH Pounder         7.9       760505847     Avatar3

Or faster solution:

nofilms =  df.groupby(["Actor_1"])['Movie_Title'].transform('size')
df1 = df[nofilms > 3]
print (df1)
        Director      Actor_1  IMDB_Score  Gross_Earnings Movie_Title
0  James Cameron  CCH Pounder         7.9       760505847      Avatar
1  James Cameron  CCH Pounder         7.9       760505847     Avatar1
2  James Cameron  CCH Pounder         7.9       760505847     Avatar2
3  James Cameron  CCH Pounder         7.9       760505847     Avatar3

Then use groupby and aggregate mean:

df2 = df1.groupby('Actor_1')['Gross_Earnings'].mean()
print (df2)
Actor_1
CCH Pounder    760505847
Name: Gross_Earnings, dtype: int64

And last plot by Series.plot.barh:

df2.plot.barh()

Upvotes: 1

Related Questions