1 Table of Contents

• Demand
• Intensity
• Heat map
• Bin plots

2 Getting started

UConn EnergyStat’s high-frequency dashboard offers a colorful range of options to visualize the 15-minute utility consumption data. Each tab visualizes the same data in a different way. For example, Demand and Intensity create line charts to demonstrate the historical trend of energy use; Heat map allows comparing consumption of a time period throughout the week; and Bin plots shows the number of observations for a variable, represented by the height of each bin.

3 Demand

Demand generates line charts of all available services for one specific building. In each plot, the line representing utility consumption is superimposed by lines conveying weather information such as temperature and relative humidity. The major difference between Demand and Intensity is that Demand shows unprocessed native consumptions whereas Intensity standardizes the native consumptions by the size of the building to allow comparison between buildings of varying sizes.

4 Intensity

Intensity is similar to Demand except that the consumptions are divided by the size of the building, since it is natural for a bigger building to consume more than a smaller equivalent. This in turn enables comparing consumptions between two or more buildings.

The specifics for Intensity are largely identical to those of Demand.

1. Click on the service_tag field to select the utility use with the desired unit.

2. Set the start_date and end_date fields according to your needs.

3. Click on the buildings field to select one or more buildings.

SCIENCE button provides quick access to five science buildings: Biology/Physics, Chemistry Building, Pharmacy/Biology Building, AG Bio-Technology, and Advanced Technology Laboratory.

4. Click on the submit button to generate plots.

5 Heat map

Heat maps serve a similar purpose to Intensity’s: compare the intensity of the utility consumption. If Intensity allowed comparing between buildings, however, Heat map rather allows comparing between days in a given week. Every 15-minute interval is represented by a color-coded cell in the heat map. If the corresponding consumption is bigger relative to the selected week’s overall consumption, it will be more red than other cells. Conversely, a consumption relatively low will be color-coded blue. This helps identify the day of the week where the most consumption took place.

1. START DATE and END DATE are the same as Demand and Intensity—see previous sections. WEEKS AVAILABLE field finds full weeks around those dates. If the set dates aren’t divisible into full weeks, HEAT MAP will automatically pad the missing days by locating the nearest Sunday before the START DATE and the nearest Saturday after the END DATE.

2. Choose a building.

3. Choose a service.

4. Click on the submit button to generate the heat map.

6 Bin plots

Bin plots help visualize the distribution of a particular variable. Currently, the Bin plots mirrors a sister website developed in R shiny. This web page will update dynamically as the fields are updated, unlike other tabs. Therefore, generating the plots are relatively straightforward.

• The first row shows the the summary statistic—as set in the summary field—for the range of temperature of the corresponding month.

• The second row shows the number of hours that recorded a temperature in the corresponding temperature range.

• Next set of bin plots visualizes the same data slightly differently—the x-axis and the color of each bar are reversed. Bars represent the service usage (or number of hours) for the temperature that corresponds to the color.

• Subsequent bin plot visualizes the interquartile range (IQR) divided by the median of the data corresponding to the month in the x-axis and the temperature range represented by the bar’s color. The IQR captures the fluctuation in the data set, which is not invariant to the measurement unit—e.g., for the same data, IQR in miles is different from the IQR in feet. To remove the effect of the measurement unit, median, representing the center of the data, is used to divide the IQR. The resulting IQR/med becomes a unit-invariant (standardized) statistic of a data set’s dispersion.