Some Pandas handy snippets for Data Scientists

Iterate through rows

import pandas as pd

for index, row in df.iterrows():
    pass

Count unique values in dataframe

df.labels.value_counts()

Style max value in a row or column

# Inspired by https://stackoverflow.com/a/45606572/2874139
def highlight_max(data, color='yellow', isBold=True):
    # Styling
    attrs = []
    if color is not None:
        attrs.append(f'background-color: {color}')
    if isBold:
        attrs.append('font-weight: bold')
    attrs = '; '.join(attrs)

    if data.ndim == 1:
        is_max = data == data.max()
        return [attrs if value else '' for value in is_max]
    else:
        is_max = data == data.max().max()
        return pd.DataFrame(np.where(is_max, attrs, ''), index=data.index, columns=data.columns)

df.style.apply(highlight_max, axis=1) # Max in row
df.style.apply(highlight_max, axis=0) # Max in column

Display 1000 rows and columns

# source: fast.ai material
def display_all(df):
    with pd.option_context("display.max_rows", 1000, "display.max_columns", 1000): 
        display(df)

display_all(df)

Save dataframe as CSV file

# index specifies whether to add a sequential index to the saved file
df.to_csv(csv_path, index=False)

Create dataframe form python dictionary

all_questions = []  # rows of column 'all_questions'
all_good_answers = []  # rows of column 'all_good_answers'
all_bad_answers = []  # rows of column 'all_bad_answers'

qa_dict = {'question': all_questions, 'good_answer': all_good_answers, 'bad_answer': all_bad_answers}

# Create a dataframe with 3 columns: question, good_answer and bad_answer
df = pd.DataFrame(data=qa_dict)

Parse dates in dataframe

df = pd.read_csv("train.csv", low_memory=False, parse_dates=["createddate"])

---

I am working on a project called ML Studio, want to get early access to and product updates? Subscribe here or follow me on twitter.