Electricity Usage Visualization

by Fan Hong and Daye Chen

To visualize the electricity usage of ~40 commercial users in Ningxia Province.

  • Visualize electricity usage behaviors of commercial usage;
  • Find relationship between places and electricity usage;
  • Find relationship between industries and electricity usage;
  • Fidn relationship between periods and electricity usage.

Electricity usage data of 40 commercial users in Ningxia Province

Sample rate: every 15 minites, 96 samples per day;

Time range: different for different users, the widest range is from June 2010 to March 2012;

Data characteristics: Sampling is not continuous, the ranges of values are not uniform.

Weather data

weather, highest and lowest temperature. But we do not use.

Location data for 40 commercial users

Latitude and longitude(collected from Google map).

Boundaries data for cities in Ningxia Province

Polygons enclosed by points of latitude and longtitude(collected from Google map).


To select points representing commercial users in the map, then to visualize and analyse usage data in barcharts and piecharts, and to adjust parameters for different purposes.



Political map: A brief map to show cities in Ningxia Province. Click on one city will lead the Google map zooming & panning to corresponding area.

Google map: Zoom & pan interaction is supported. Points are overlaid on the map with accurate positions which is used to represen commercial users.

Overview Select city Zoom & Pan


The whole view contains multiple barcharts each of which represents one commercial users. And barcharts(users) are organized as a hierarchical structure.

As for the single barchart, two parameters, granularity and period, affects the visualzation.

The whole x axis is used to represent the time range of selected period. If period parameter is set to “one week”, x axis will represent one week; if period parameter is set to “None”, x axis will represent the whole time range of data. And when period parameter is not set to “None”, the data is also aggregated from the time intervals in different periods. The meaning of one single bar correspond to granularity parameter. If granularity parameter is set to “one hour”, a bar is used to show values of one hour. A single bar is used to show mean value, maximum value and minimum value of that place in that time interval.

And for the hierarchical structure of multiple barcharts(users), we can group barcharts(users) by dragging one barchart to another. Clicking a non-leaf node will “zip” that subtree into a single barchart to represent the aggregation of users in that subtree.

Also, if we drag the barchart out of the right boundary, the selected barchart(user) will be eliminate from visualzation.

Group barcharts(users) “zip” subtree


It also contains multiple piecharts organized as a hierarchical structure.

Granularity and period parameters affects piechart view, but in different ways. A ring represents one period, and multiple rings of different periods will be stacked along the radial direnction whose radii increase with time. A ring is assembled with multiple arc which corresponds to mean value of one time interval of selected granularity.

For the hierarchical structure, interactions in barcharts are still valid in piecharts.

Widgets for adjusting parameters

Besides adjusting granularity and period, we can also change the time range of data we use. This will filter all usage data of all selected users and grouped users ahead of visualization. And because the value ranges of different users varies a lot, this will lead to different scales in y axis of barcharts. Therefore, we can choose whether to use uniform value range for all selected users or not.


  • Choose a city and zoom & pan areas in the map;
  • Select commercial user(s) in the map;
  • Adjust parameters (granularity, period, time range, etc.) for barcharts and piecharts;
  • Group and integrate users;
  • Eliminate user(s).

We can learn a lot from the visualization. Now there are some examples of the conclusion.

Case 1

The electricity usage of department stores in different days of one week.

From these four charts, we can learn the electricity usage of department stores is stable during a week.

Case 2

The electricity usage of restaurants in different hours of one day.

From these charts, we can learn the electricity usage of the last 3 restaurants are in a high level during lunch time (12:00-13:00) and evening time (19:00 - 23:00) and in a low level during the rest of day.

However, the first user has a different usage pattern. It has high level usage from 20:00-6:00, but a relatively low level usage during day time. We can give more attention to this user.

Case 3

The electricity usage of hotels and department shops in different hours of one day.

From these charts, we can learn the electricity usage of restaurants and departments has different climax time and duration. So the power supply company can adjust the time of power supply and objects.