Pandas 常用方法小结

 基本数据读取存储

import pandas as pd

filename = "123.csv"
save_filename = "save_123.csv"

# basic usage
## load csv file
data = pd.read_csv(filename, encoding="utf-8")

## save to another file
data.to_csv(save_filename, mode="w", encoding="utf-8", index=False)

## from dict to dataframe
tmp_dict = {"a":["1","2","3", "2"], \
		    "b":[2, 3, 4, 5], \
		    "c":["123","234","345","235"]
		   }
df = pd.DataFrame(tmp_dict, columns = ["a", "b", "c"])

## print the first xx row
print(df.head())

# get index 
indexes = df.index.tolist()

## iterate rows
for idx, row in df.iterrows():
    tmp_a = row["a"]
    tmp_b = row["b"]
    print(tmp_a, tmp_b)

## add/modify value in row 2
idx = 2
df.at[2, "b"] = 4
df.at[5, "b"] = 5

## remove certain rows
df_drop = df.drop([1,2])

# get certain row
certain_row = df.iloc[3]

## get certain column 
a_column = df["a"].tolist()

## count frequency
df.a.value_counts()

## sort values
df_sorted = df.sort_values(["b"], ascending=[True])

## random sample 
df_sample = df.sample(n=3)

## get unique values for certain column
df_unique_a = df["a"].unique()

## group data by certain column, 
## and calculate the number of unique b in each group
df_group = df.groupby("a")["b"].nunique()

各种过滤

# apply filter
## filter rows whose a == "1"
df_no_1 = df.loc[df["a"] == "1"]

## filter rows whose b > 2
df_larger_than_2 = df[df["b"]>2]

## filter rows whose c contain "2"
df_has_2 = df.loc[df["c"].str.contains("2")]

## filter rows wh