

In [1]:

    
%matplotlib inline

BLS Timeseries Data Exploration

In this workbook, I've set up a data frame of Bureau of Labor Statistics time series data, your goal is to explore and visualize the time series data using pandas, matplotlib, seaborn, or even Bokeh!



In [9]:

    
# Imports 
import csv 
import numpy as np 
import pandas as pd
import seaborn as sns 
import matplotlib.pyplot as plt 

from itertools import groupby
from operator import itemgetter

Data Loading

The data is stored in a zip file in the data directory called data/bls.zip -- unzip this file and there are two CSV files. The first series.csv is a description of the various time series that are in the data frame. The second complete record of the time series data, with the associated time series id.

We we load the series information into it's own dataframe for quick lookup (like a database) and then create a dataframe of each individual series data, identified by their ID. There is more information in the CSV, which you can explore if you'd like.



In [4]:

    
# Load the series data 
info = pd.read_csv('../data/bls/series.csv')

def series_info(blsid, info=info):
    return info[info.blsid == blsid]

# Use this function to lookup specific BLS series info. 
series_info("LNS14000025")









    Out[4]:






  
    
      
      id
      blsid
      title
      source
      is_primary
      delta_id
      is_delta
      is_adjusted
    
  
  
    
      20
      1635
      LNS14000025
      Unemployment Rate - 20 Years & Over Men
      CPS
      t
      1708.0
      f
      f



In [ ]:

    
# Load each series, grouping by BLS ID
def load_series_records(path='../data/bls/records.csv'):
    with open(path, 'r') as f:
        reader = csv.DictReader(f)
        
        for blsid, rows in groupby(reader, itemgetter('blsid')):
            # Read all the data from the file and sort 
            rows = list(rows) 
            rows.sort(key=itemgetter('period'))
            
            # Extract specific data from each row, namely:
            # The period at the month granularity 
            # The value as a float 
            periods = [pd.Period(row['period']).asfreq('M') for row in rows]
            values = [float(row['value']) for row in rows]
            
            yield pd.Series(values, index=periods, name=blsid)
            

series = pd.concat(list(load_series_records()), axis=1)
series