Energy Consumption Prediction in the home appliances

Base paper link: https://doi.org/10.1016/j.enbuild.2017.01.083
Datasets Details: https://www.kaggle.com/loveall/appliances-energy-prediction

Note: Read the base paper and below listed points before starting the work.

3. No of pages:
We shall consider five – A4 double spaced pages with Aerial 12point, 6th pages we shall use for the references detail & 7th page we shall use for the Time plan

4. Introduction: (General Suggestion from Tutor)
a. The introduction has to start with the purpose of this research.
b. How energy is used in Belgium and home appliances usage details since the dataset is from the Belgium country. (Since the selected dataset is from the Belgium country)
c. In addition, the research how will help Belgium policymakers and others.
d. Further information with references (minimum 3-4 ).

5. Aim:
The aim is the prediction of energy consumption from home appliances, namely televisions, rice cookers, electric water heaters, computers, washing machines, and refrigerators.

6. Objectives: (below-listed objectives are accepted by the tutor, further to be developed with 3more points)
a. Pre-processing the dataset.
b. Access the best statistics detail from the dataset. (like peak energy consumption at daily, weekly & monthly, what variables make the consumption high & low, when occupant changes what impacts in the energy consumption)
c. To compute the ‘error rate of prediction’ for ‘enhancing the prediction of energy on appliances results’
d. Future energy prediction for half-day, day, month & year.

7. Background: (General Suggestion from Tutor)
a. The dataset details to be covered.
b. Minimum 4-5 references to be used in the background, the reference 1 was using the ‘X’ technique to achieve this; reference 2 was using the ‘Y’ technique and others. No need to write so many details about reference research.
c. If someone used the same dataset to do the research with different models then have to explain in very short about that and how this going to differ. (Here with the same dataset someone used for his research work, that’s the base paper for my research)

8. Methodology: (General Suggestion from Tutor)
a. What is in these datasets? What does the data look like? What pre-processing steps need to be taken? How are the ML techniques you propose to use going to be used? Why are they relevant?.
b. From the datasets how we are going to do the pre-processing like addressing the non-recorded values from the dataset, rearranging all records with the same length, and others.
c. Best suitable method to achieve the aim & objectives with formulas and others.
d. The steps need to be related to the plan of work and the processing of the data
e. Some figures about methods.
f. Many references to be added.

9. Time plan: (The time plan should be on a weekly basis )
a. Has to cover the complete task. (like pre-processing is one task, error rate finding is one task, statistical development is one task likewise what are the task involved all to be covered in the time plan)

10. Dataset: If you want to change the dataset, feel free to do that whichever is matching to do the home appliances energy prediction research. (Since I cannot the topic name)

11. References: Harvard reference style to be followed.

 

12. Brief on Dataset:.
1) The datasets contain Temperature (T in deg. C), Relative Humidity (RH in %), appliances’ energy usage (Wh), Lights energy usage (Wh), and weather data reports from the nearest airport.
2) Watt-hour(Wh): If a 60 W light bulb is on for one hour, then that light bulb will have used 60 Wh of energy. If left on for two hours, then the 60 W light bulb will have used 120Wh of energy.
3) The Temperature (T) & Relative Humidity (RH) were measured from all rooms, outside room & living room through ZigBee wireless network Sensor(WNS).
4) The energy usage is measured from the M-Bus energy counters, like how we have an electricity meter in our houses, the same way they used meters to record the appliance’s energy usage & lighting energy usage separately.
5) The weblink to the dataset should either be a reference to a report or added as a footnote