https://github.com/hebes-io/eensight

The measurement and verification methodology of the H2020 project SENSEI
https://github.com/hebes-io/eensight

building-energy energy-data energy-efficiency kedro pipelines

Last synced: 3 months ago
JSON representation

The measurement and verification methodology of the H2020 project SENSEI

• Host: GitHub
• URL: https://github.com/hebes-io/eensight
• Owner: hebes-io
• License: apache-2.0
• Created: 2020-05-16T16:56:19.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2023-01-22T11:56:03.000Z (almost 2 years ago)
• Last Synced: 2024-03-22T04:38:44.292Z (8 months ago)
• Topics: building-energy, energy-data, energy-efficiency, kedro, pipelines
• Language: Python
• Homepage: https://hebes-io.github.io/rethinking/
• Size: 37.7 MB
• Stars: 16
• Watchers: 4
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• open-sustainable-technology - eensight - This Python package implements the measurement and verification (M&V) methodology that has been developed by the H2020 project SENSEI - Smart Energy Services to Improve the Energy Efficiency of the European Building Stock. (Consumption / Buildings and Heating)

README

        
![logo](https://github.com/hebes-io/eensight/blob/master/logo.png)





[![PyPI version](https://badge.fury.io/py/eensight.svg)](https://badge.fury.io/py/eensight)

## The `eensight` tool for measurement and verification of energy efficiency improvements

The `eensight` Python package implements the measurement and verification (M&V) methodology that has been developed by the H2020 project [SENSEI - Smart Energy Services to Improve the Energy Efficiency of the European Building Stock](https://senseih2020.eu/). 

The online book *Rethinking Measurement and Verification of Energy Savings* (accessible [here](https://hebes-io.github.io/rethinking/index.html)) explains in detail both the methodology and its implementation.

## Installation

`eensight` can be installed by pip:

```bash

pip install eensight

```

## Usage

### 1. Through the command line

All the functionality in `eensight` is organized around data pipelines. Each pipeline consumes data and other artifacts (such as models) produced by a previous pipeline, and produces new data and artifacts for its successor pipelines.

There are four (4) pipelines in `eensight`. The names of the pipelines and the associations between pipelines and namespaces are summarized below:

|            	| train    	| test   	| apply   |

|------------	|----------	|----------	|---------|

| preprocess 	| ✔ 	| ✔ 	| ✔|

| predict    	| ✔ 	| ✔	| ✔|

| evaluate    	|          	| ✔  | ✔|

| adjust    	|          	|           | ✔|

The primary way of using `eensight` is through the command line. The first argument is always the name of the pipeline to run, such as:

```bash

eensight run predict --namespace train

```

The command

```bash

eensight run --help

```

prints the documentation for all the options that can be passed to the command line.

### 2. As a library

The pipelines of `eensight` are separate from the methods that implement them, so that the latter can be used directly:

```python

import pandas as pd

from eensight.methods.prediction.baseline import UsagePredictor

from eensight.methods.prediction.activity import estimate_activity

non_occ_features = ["temperature", "dew point temperature"]

activity = estimate_activity(

    X, 

    y, 

    non_occ_features=non_occ_features, 

    exog="temperature",

    assume_hurdle=False,

)

X_act = pd.concat([X, activity.to_frame("activity")], axis=1)

model = UsagePredictor(skip_calendar=True).fit(X_act, y)

```