代做Modern Time Series Assignment调试R语言程序

Modern Time Series Assignment

Instruction

● Select a free time-series dataset. If you do not know where to find a free time-series dataset, "Anomaly Detection for Time Series" has listed a few resources for you to start with.

● Perform. the following procedures in your work:

○ Simple moving average and Exponential smoothing (using Pandas and Matplotlib)

○ Seasonal-Trend Decomposition (using Prophet or NeuralProphet)

● Additional Instructions:

○ Please justify why each of your models can identify anomalies.

○ For simple moving average and exponential smoothing, visual inspection is fine but be sure to justify your reasoning through use of histograms, winsorizing or other techniques discussed in week 1 (review Basic-Detecting-Anomalies notebook).

○ For seasonal-trend decomposition please review the notebook examples discussed in class and perform. a similar analysis on your dataset. Try to model the anomaly as a custom holiday (review notebook Prophet-Detecting-Anomalies). Create a forecast (months/years into future is your choice) and apply the anomaly somewhere in the future. Comment on the impact of the anomaly in your forecast.

● Submit in the HTML format.

Assessment

Homework grading: please strive to achieve the best score, although in the past 19-20 is rare.

● 20-18 exceed expectation.

○ You have performed the models well. You demonstrate an expert ability to communicate your knowledge

● 18-14 meets expectations

○ You have performed the models, yet some technical decisions should have been considered to make your analysis better

● 14-12 nearly meets the expectation

○ Although you have performed the basic exploratory data analysis, your technical decisions are flawed. And your documentation for the insights should be improved.

● 12 - do not meet the expectation

○ Although you have performed the basic modeling tasks, your models are incorrect, and your analysis is unsatisfactory.

● Late <=1 day: -1

● Late >1 day and <=2 days: -2

● Late >2 days: -5

Submission

1. Submit the report as HTML and the code as the notebook source.

2. Click the blue Submit Assignment button at the top of this page.

3. Click the Choose File button, and locate your submission.