pd.read_csv() by default returns a DataFrame 📄 (2D table).
✅ What if you want a Series instead of a DataFrame?
🟣 Case 1: CSV has only one column
If your CSV file has just one column of data (not including index), you can use:
python
CopyEdit
series = pd.read_csv("file.csv", squeeze=True)
✅ squeeze=True converts single-column DataFrame → Series
⚠️But squeeze is deprecated, so better method:
python
CopyEdit
df = pd.read_csv("file.csv")
series = df.iloc[:, 0] # Select 1st column as Series
🟣 Case 2: CSV has multiple columns, but you want only one as Series
Suppose your file has 5 columns: ['Name', 'Age', 'City', 'Marks', 'Grade']
You can load only one column as Series like this:
python
CopyEdit
# Read only one column
series = pd.read_csv("file.csv", usecols=['Marks']).squeeze()
✅ usecols picks only the required column
✅ .squeeze() converts it from DataFrame → Series
🟡 If you want multiple columns but each as a separate Series:
python
CopyEdit
df = pd.read_csv("file.csv")
name_series = df['Name']
age_series = df['Age']
Each column is still a Series.
You can access any column in a DataFrame like this:
python
CopyEdit
df['column_name'] # returns a Series
✨ Summary:
Situation Code Returns
CSV has 1 column pd.read_csv("file.csv") DataFrame
pd.read_csv(...).squeeze() or df.iloc[:,
Force Series 0]
Series
One column from pd.read_csv(..., usecols=[...]).squeeze() Series
many
Multiple columns Use df['col'] for each Series (one at a time)
✅ What [-1] means in Python?
In regular Python lists or NumPy arrays, [-1] means:
“Give me the last item.”
Example:
python
CopyEdit
lst = [10, 20, 30]
print(lst[-1]) # Output: 30
🔍 In Pandas, this behaves differently depending on the type of object:
✅ Case 1: Series[-1]
If runs is a Pandas Series, then doing:
python
CopyEdit
runs[-1]
...is not the last element. It tries to access the index label -1, not the position.
🔸 If -1 is not an index label in runs, it gives an error:
python
CopyEdit
KeyError: -1
🧠 Solution: Use .iloc[-1] for position-based access
python
CopyEdit
runs.iloc[-1] # ✅ Last row by position
✅ Case 2: Series with numeric index including -1
python
CopyEdit
import pandas as pd
s = pd.Series([100, 200, 300], index=[0, 1, -1])
print(s[-1]) # ✅ Works! Because -1 is an index label
So, [-1] works if your index contains -1 as a label. Otherwise, it fails.
✅ Case 3: Why movies[-1] worked?
Possibilities:
1. movies might be a Python list or NumPy array → so [-1] gives last item
2. Or maybe movies is a Series with -1 as a valid index
🔑 Golden Rule in Pandas:
Task Use
Access last item by position series.iloc[-1] ✅
Access last item by label (if label is -1) series[-1]
Access last item in DataFrame df.iloc[-1]
✨ Summary
Expression Works if... Preferred?
series[-1] Only if -1 is in index ❌ Risky
Expression Works if... Preferred?
series.iloc[-1] Always works by position ✅ Best
list[-1] Always works (Python list) ✅
🎯 What is category dtype?
It’s a special data type in pandas used for columns with repeated values, like:
python
CopyEdit
['Male', 'Female', 'Male', 'Female', 'Other', 'Female'...]
Instead of storing full strings every time, pandas stores codes (integers) behind the scenes.
🧠 Example:
python
CopyEdit
import pandas as pd
df = pd.DataFrame({
'gender': ['Male', 'Female', 'Male', 'Female', 'Other']
})
df['gender'] = df['gender'].astype('category')
Now df['gender'] is stored as:
Value Code
Male 0
Female 1
Other 2
Internally, it maps the string to small integers → uses less memory.
✅ Benefits:
Feature Benefit
Memory saving ✅ Huge for large datasets
Feature Benefit
Speed ✅ Faster for filters, groupby
Sorting ✅ Can be more efficient
Predefined categories ✅ Use .cat.categories
📊 When should you use it?
Use category when:
The column has repeated string values (low unique %)
You want to save memory
You’re doing operations like groupby, merge, or filter
That’s okay, Nitin! Let me explain it clearly and step by step. You're asking:
Great catch, Nitin! 👏 You're seeing this warning:
DeprecationWarning: DataFrameGroupBy.apply operated on the grouping columns. This
behavior is deprecated...
Let’s break this down so it’s super clear:
⚠️What's Happening?
In older Pandas versions, when you do:
python
CopyEdit
grouped = genres.groupby('Genre')
grouped.apply(foo)
Your function foo() gets a DataFrame that includes the grouping column ('Genre' in this
case).
But in newer Pandas (from v2.0 onwards), this will be deprecated, meaning the grouping
column won't be included by default. That’s why you’re getting the warning.
✅ Fix 1: Use include_group=False (Recommended way)
If your function doesn’t need the group key, write:
python
CopyEdit
grouped.apply(foo, include_group=False)
This silences the warning and future-proofs your code.
Can I use axis=0 in groupby.apply() to fix the warning?
❌ Short Answer: No, axis=0 does not solve this warning.
💡 Why?
Because:
groupby.apply() works group by group, not row by row or column by column.
axis=0 or axis=1 is not even an argument for groupby.apply() — it's used in
functions like DataFrame.apply(), not groupby.
So this line gives a warning:
python
CopyEdit
genres.groupby('Genre').apply(foo)
And axis=0 won't help:
python
CopyEdit
genres.groupby('Genre').apply(foo, axis=0) ❌ ERROR
✅ How to fix the warning?
You need to write either:
✅ Option 1: Add include_group=False
python
CopyEdit
genres.groupby('Genre').apply(foo, include_group=False)
This tells Pandas:
"Don't pass the group column (like 'Genre') to the foo function."
No warning. Future-safe.
✅ Option 2: Select only numeric columns before apply()
python
CopyEdit
genres.groupby('Genre')[['IMDB_Rating', 'Gross']].apply(foo)
This avoids including the group column ('Genre'), which also silences the warning.
🔁 When can you use axis=0?
You can use axis=0 in:
DataFrame.apply() → operates column-wise
DataFrame.apply(func, axis=1) → operates row-wise
✅ Example:
python
CopyEdit
df.apply(np.min, axis=0) # column-wise min
df.apply(lambda row: row.sum(), axis=1) # row-wise sum
But not with .groupby().apply().
🧠 Summary
Use Case Can use axis? Best Fix
df.apply() ✅ Yes Use axis=0 for columns, axis=1 for rows
Use include_group=False or select columns
df.groupby(...).apply() ❌ No
manually
By default, Pandas sends the entire group including the 'Genre' column (used for grouping)
into your function foo.
But in the future, Pandas will remove that by default — so it gives a warning now.
⚙️What include_group=False Does
It tells Pandas not to include the grouping column (like 'Genre') when calling your function
on each group.
So instead of getting this inside foo:
nginx
CopyEdit
Genre IMDB_Rating Gross
0 Drama 9.0 200M
1 Drama 8.5 150M
You’ll only get:
nginx
CopyEdit
IMDB_Rating Gross
0 9.0 200M
1 8.5 150M
This is cleaner and avoids accidental bugs.
✅ Benefit
Silences the deprecation warning
Future-safe for Pandas 2.x+
Helps your function avoid unexpected behavior from extra columns
✨ Real Example
python
CopyEdit
def foo(group):
return pd.Series({
'Avg_Rating': group['IMDB_Rating'].mean(),
'Total_Gross': group['Gross'].sum()
})
result = df.groupby('Genre').apply(foo, include_group=False)
This runs without warnings ✅
Your function foo() now gets only the columns it needs.
🧠 Summary
With include_group=True (default) With include_group=False
Your function gets 'Genre' + data Your function gets only data
Can cause warnings No warning ✅
Might break in future Future-safe ✅
You wrote this function:
python
CopyEdit
def foo(group):
return group['Series_Title'].str.startswith('A').sum()
And you're applying it:
python
CopyEdit
df.groupby('Genre').apply(foo)
This works and gives you the number of movies starting with "A" in each genre.
🧠 Now the Question:
"Even though 'Genre' column is passed, I’m not using it inside the function. So why does it
warn me?"
🤔 Why the Warning?
Because:
Pandas is still including the 'Genre' column (used for grouping) in the DataFrame
that it passes to your foo() function.
Even if you don't use the 'Genre' column, it’s still there in the input.
Pandas wants to change this behavior in future versions, so it warns you now to
future-proof your code.
🔎 What Actually Happens Now (Pandas 1.x and 2.x):
Inside your function, for each group like 'Action', you're getting this:
python
CopyEdit
Genre Series_Title
0 Action Avengers
1 Action Aquaman
2 Action Avatar
You just use 'Series_Title', but 'Genre' is still present in that group.
✅ How to silence the warning?
Add this:
python
CopyEdit
df.groupby('Genre').apply(foo, include_group=False)
Now, inside foo(), the group will look like this:
python
CopyEdit
Series_Title
0 Avengers
1 Aquaman
2 Avatar
No 'Genre' column is passed.
📝 Summary
Should you fix
Code What happens
it?
Works, but gives
df.groupby('Genre').apply(foo) ✅ Yes
warning ⚠️
df.groupby('Genre').apply(foo, Same result, no
include_group=False) ✅ Best
warning ✅
✅ Clean, but
df.groupby('Genre')[['Series_Title']].apply(foo) Also works
manual
✅ Final Code (Clean and Future-Safe)
python
CopyEdit
def foo(group):
return group['Series_Title'].str.startswith('A').sum()
result = df.groupby('Genre').apply(foo, include_group=False)
Let me know if you want to count movies starting with other letters too (like 'T' or vowels) — I
can make it dynamic 😊
