Getting Started With MapD, Part 2: Electricity Dataset

In my previous MapD post, I loaded electricity data into MapD Community Edition, intentionally ignoring the what of the data to keep that post from being too overwhelming. Now let’s take a step back and explain the dataset, show how to format the data using Python that was loaded into MapD, then use the MapD Immerse UI to build a simple dashboard.

PJM Metered Load Data

I started off my career at PJM doing long-term electricity demand forecasting, to help power engineers do transmission line studies for reliability and to support expansion of the electrical grid in the U.S. Because PJM is a quasi-government agency, they provide over 25 years of hourly electricity usage for the Eastern and Central U.S., both in aggregate and by local power region (roughly, the local power company territories).

However, just because the data is available doesn’t mean it’s convenient, and unfortunately, the data are stored as Excel spreadsheets. This is easily remedied using pandas (v0.22.0, python3.6):

import os
import pandas as pd

#change to directory with files for convenience

#first sheet in workbook contains all info for years 1993-1999
df1993_1999 = [pd.read_excel(str(x) + "-hourly-loads.xls") for x in range(1993,1999)]

#melt, append df1993-df1999 together
df_melted = pd.DataFrame()
for x in df1993_1999:
    x.columns = df1993_1999[1].columns.tolist()
    x_melt = pd.melt(x, id_vars=['ACTUAL_DATE', 'ZONE_NAME'], var_name = "HOUR_ENDING", value_name = "MW")
    df_melted = df_melted.append(x_melt)

#multiple sheets to concatenate
#too much variation for a one-liner
d2000 = pd.read_excel("2000-hourly-loads.xls", sheet_name = [x for x in range(2,17)])
d2001 = pd.read_excel("2001-hourly-loads.xls", sheet_name = None)
d2002 = pd.read_excel("2002-hourly-loads.xls", sheet_name = [x for x in range(1,18)])
d2003 = pd.read_excel("2003-hourly-loads.xls", sheet_name = [x for x in range(1,19)])
d2004 = pd.read_excel("2004-hourly-loads.xls", sheet_name = [x for x in range(2,24)])
d2005 = pd.read_excel("2005-hourly-loads.xls", sheet_name = [x for x in range(2,27)])
d2006 = pd.read_excel("2006-hourly-loads.xls", sheet_name = [x for x in range(3,29)])
d2007 = pd.read_excel("2007-hourly-loads.xls", sheet_name = [x for x in range(3,29)])
d2008 = pd.read_excel("2008-hourly-loads.xls", sheet_name = [x for x in range(3,29)])
d2009 = pd.read_excel("2009-hourly-loads.xls", sheet_name = [x for x in range(3,29)])
d2010 = pd.read_excel("2010-hourly-loads.xls", sheet_name = [x for x in range(3,29)])
d2011 = pd.read_excel("2011-hourly-loads.xls", sheet_name = [x for x in range(3,33)])
d2012 = pd.read_excel("2012-hourly-loads.xls", sheet_name = [x for x in range(3,33)])
d2013 = pd.read_excel("2013-hourly-loads.xls", sheet_name = [x for x in range(3,34)])
d2014 = pd.read_excel("2014-hourly-loads.xls", sheet_name = [x for x in range(3,34)])
d2015 = pd.read_excel("2015-hourly-loads.xls", sheet_name = [x for x in range(3,40)])
d2016 = pd.read_excel("2016-hourly-loads.xls", sheet_name = [x for x in range(3,40)])
d2017 = pd.read_excel("2017-hourly-loads.xls", sheet_name = [x for x in range(3,42)])
d2018 = pd.read_excel("2018-hourly-loads.xls", sheet_name = [x for x in range(3,40)])

#loop over dataframes, read in matrix-formatted data, melt to normalized form
for l in [d2000, d2001, d2002, d2003, d2004, d2005, d2006, d2007, d2008, d2009, d2010,
          d2011, d2012, d2013, d2014, d2015,d2016,d2017,d2018]:

    for k in l:
        temp = l[k].iloc[:, 0:26]
        temp.columns = df1993_1999[1].columns.tolist()
        df_melted = df_melted.append(pd.melt(temp, id_vars=['ACTUAL_DATE', 'ZONE_NAME'], var_name = "HOUR_ENDING", value_name = "MW"))

#(4898472, 4)
#152MB as CSV
df_melted.to_csv("hourly_loads.csv", index=False)

The code is a bit verbose, if only because I didn’t want to spend time to figure out how to programmatically determine how many tabs each workbook has. But the concept is the same each time: read an Excel file, get the data into a dataframe, then convert the data to long form. So instead of having 26 columns (Date, Zone, Hr1-Hr24), we have 4 columns, which is quite frequently a more convenient way to access the data (especially when using SQL).

The final statement writes out a CSV of approximately 4MM rows, the same dataset that was loaded using mapdql in the first post.

Top 10 Usage Days By Season

One of the metrics I used to monitor as part of my job was the top 5/top 10 peak electricity use days per Summer (high A/C usage) and Winter (electric space heating) seasons. Back in those days, I used to use SAS against an enterprise database and the results would come back eventually

Obviously, it’s not a fair comparison to compare today’s GPUs vs. late ’90s enterprise databases in terms of performance, but back then it did take a non-trivial amount of effort to run this query to keep the report updated. With MapD, I can do the same report in ~100ms:

--MapD doesn't currently support window functions, so need to precalculate maximum by day
with qry as (select
max(MW) as daily_max_usage
from hourly_loads
where zone_name = 'MIDATL' and actual_date between '2017-06-01' and '2017-09-30'
group by 1,2)
from hourly_loads as hl
inner join qry on qry.actual_date = hl.actual_date and qry.daily_max_usage =
order by daily_max_usage desc
limit 10;

top 10 electric usage

The thing about returning an answer in 100ms or so is that its fast enough where calling these results from a webpage/dashboard would be very responsive; that’s where MapD Immerse comes in.

Building A Dashboard Using MapD Immerse

Rather than copy/pasting the query in and running it, it’s pretty easy to build an automated report using the Immerse dashboard builder. I’m limited to a single data source because I’m using MapD Community Edition, but in just a few minutes I was able to create the following dashboard:

mapd immerse dashboard

I took the query from above and built a view to encapsulate the query, so I didn’t have to worry about the with statement or joins, I could just use the view as if the results were pre-calculated. From there, adding in a results table and two bar charts was fairly quick (in the same drag-and-drop style of Tableau or other BI/reporting tools).

While this dashboard is pretty rudimentary in its design, were this data source set up as real-time using Apache Kafka or similar, this chart would always be up-to-date for use on a TV screen or as a browser bookmark without any additional data or web engineering.

Obviously, many dashboarding tools exist, but its important to note that no pre-aggregation or column indexing or other standard database performance tricks are being employed (outside of specialized hardware and fast GPU RAM caching). Even with 10 dashboard tiles updating serially 100ms at a time, you are still in the 1-2s page load time, on par with the fastest-loading dynamic webpages on the internet.

Programmatic analytics using pymapd

While dashboarding can be very effective for keeping senior management up-to-date, the real value of data is unlocked with more in-depth analytics and segmentation. In my next blog post, I’ll cover how to access MapD using pymapd in Python, doing more advanced visualizations and maybe even some machine learning…

  • Using RSiteCatalyst With Microsoft PowerBI Desktop
  • RSiteCatalyst Version 1.4.14 Release Notes
  • RSiteCatalyst Version 1.4.13 Release Notes
  • RSiteCatalyst Version 1.4.12 (and 1.4.11) Release Notes
  • Self-Service Adobe Analytics Data Feeds!
  • RSiteCatalyst Version 1.4.10 Release Notes
  • WordPress to Jekyll: A 30x Speedup
  • Bulk Downloading Adobe Analytics Data
  • Adobe Analytics Clickstream Data Feed: Calculations and Outlier Analysis
  • Adobe: Give Credit. You DID NOT Write RSiteCatalyst.
  • RSiteCatalyst Version 1.4.8 Release Notes
  • Adobe Analytics Clickstream Data Feed: Loading To Relational Database
  • Calling RSiteCatalyst From Python
  • RSiteCatalyst Version 1.4.7 (and 1.4.6.) Release Notes
  • RSiteCatalyst Version 1.4.5 Release Notes
  • Getting Started: Adobe Analytics Clickstream Data Feed
  • RSiteCatalyst Version 1.4.4 Release Notes
  • RSiteCatalyst Version 1.4.3 Release Notes
  • RSiteCatalyst Version 1.4.2 Release Notes
  • Destroy Your Data Using Excel With This One Weird Trick!
  • RSiteCatalyst Version 1.4.1 Release Notes
  • Visualizing Website Pathing With Sankey Charts
  • Visualizing Website Structure With Network Graphs
  • RSiteCatalyst Version 1.4 Release Notes
  • Maybe I Don't Really Know R After All
  • Building JSON in R: Three Methods
  • Real-time Reporting with the Adobe Analytics API
  • RSiteCatalyst Version 1.3 Release Notes
  • Adobe Analytics Implementation Documentation in 60 Seconds
  • RSiteCatalyst Version 1.2 Release Notes
  • Clustering Search Keywords Using K-Means Clustering
  • RSiteCatalyst Version 1.1 Release Notes
  • Anomaly Detection Using The Adobe Analytics API
  • (not provided): Using R and the Google Analytics API
  • My Top 20 Least Useful Omniture Reports
  • For Maximum User Understanding, Customize the SiteCatalyst Menu
  • Effect Of Modified Bounce Rate In Google Analytics
  • Adobe Discover 3: First Impressions
  • Using Omniture SiteCatalyst Target Report To Calculate YOY growth
  • Google Analytics Individual Qualification (IQ) - Passed!
  • Google Analytics SEO reports: Not Ready For Primetime?
  • An Afternoon With Edward Tufte
  • Google Analytics Custom Variables: A Page-Level Example
  • Xchange 2011: Think Tank and Harbor Cruise
  • Google Analytics for WordPress: Two Methods
  • WordPress Stats or Google Analytics? Yes!
  • Getting Started With MapD, Part 2: Electricity Dataset
  • Getting Started With MapD, Part 1: Docker Install and Loading Data
  • Parallelizing Distance Calculations Using A GPU With CUDAnative.jl
  • Building a Data Science Workstation (2017)
  • JuliaCon 2015: Everyday Analytics and Visualization (video)
  • Vega.jl, Rebooted
  • Sessionizing Log Data Using data.table [Follow-up #2]
  • Sessionizing Log Data Using dplyr [Follow-up]
  • Sessionizing Log Data Using SQL
  • Review: Data Science at the Command Line
  • Introducing Twitter.jl
  • Code Refactoring Using Metaprogramming
  • Evaluating BreakoutDetection
  • Creating A Stacked Bar Chart in Seaborn
  • Visualizing Analytics Languages With VennEuler.jl
  • String Interpolation for Fun and Profit
  • Using Julia As A "Glue" Language
  • Five Hard-Won Lessons Using Hive
  • Using SQL Workbench with Apache Hive
  • Getting Started With Hadoop, Final: Analysis Using Hive & Pig
  • Quickly Create Dummy Variables in a Data Frame
  • Using Amazon EC2 with IPython Notebook
  • Adding Line Numbers in IPython/Jupyter Notebooks
  • Fun With Just-In-Time Compiling: Julia, Python, R and pqR
  • Getting Started Using Hadoop, Part 4: Creating Tables With Hive
  • Tabular Data I/O in Julia
  • Hadoop Streaming with Amazon Elastic MapReduce, Python and mrjob
  • A Beginner's Look at Julia
  • Getting Started Using Hadoop, Part 3: Loading Data
  • Innovation Will Never Be At The Push Of A Button
  • Getting Started Using Hadoop, Part 2: Building a Cluster
  • Getting Started Using Hadoop, Part 1: Intro
  • Instructions for Installing & Using R on Amazon EC2
  • Video: SQL Queries in R using sqldf
  • Video: Overlay Histogram in R (Normal, Density, Another Series)
  • Video: R, RStudio, Rcmdr & rattle
  • Getting Started Using R, Part 2: Rcmdr
  • Getting Started Using R, Part 1: RStudio
  • Learning R Has Really Made Me Appreciate SAS