# Forecasting a Time Series Model for 1000s of Time Series

I'm currently immersed in a challenging forecasting project centred around predicting the required work hours to complete various tasks within a team setting. My dataset comprises crucial attributes, including team IDs, task IDs, hours, and dates. Specifically, I'm working with a comprehensive dataset containing distinct time series information for each unique combination of teams and tasks, resulting in approximately 8000 distinct time series. I aim to construct a robust forecasting model tailored to this intricate scenario.

Amidst this endeavour, I've encountered several complexities. These include the diversity in time series lengths, the presence of both new and well-established teams and tasks, varying from 3 months to 2 years of data, and the potential incompleteness or gaps within the time series.

To provide deeper insight into the dataset's dynamics, each team is associated with a set of tasks, such as "call customer", "draft email", and "follow up with a client". The team members record daily time entries for these tasks, collectively contributing to the intricate web of time series data. The primary objective of my forecasting model is to predict future work hours based on historical observations, facilitating informed planning and decision-making by team leads.

Below are the first ten rows of the dataframe, sorted by Date, TeamID and TaskID

You can use the below code to generate the sample data. Note that actual data is much more noisy and also has a gap in time series

import pandas as pd
import random
from datetime import datetime, timedelta

# Create date range
start_date = datetime(2022, 1, 1)
end_date = datetime(2023, 12, 31)
date_range = pd.date_range(start=start_date, end=end_date, freq='D')

# random data for teams and tasks
teamsID = [11, 12, 13]
tasksID = [1, 2, 3]

data = []

for team in teamsID:
current_date = start_date
while current_date <= end_date:
'Date': current_date,
'Team': team,
'Hours': random.randint(1, 8)  # Random hours
})
current_date += timedelta(days=1)

# Create DataFrame
df = pd.DataFrame(data)


In my pursuit of a scalable approach, I'm exploring the following strategies to enhance forecasting accuracy:

1. Utilizing a diverse ensemble of time series models (e.g., Prophet, ARIMA) to boost forecasting precision, coupled with time series ensemble techniques for prediction aggregation (as detailed here). The resultant ensemble models would be saved for future predictions.

2. Tuning hyperparameters for individual models to optimize their predictive capabilities.

3. Conducting rigorous machine learning experiments for each ensemble model to identify the most effective models based on varying hyperparameter configurations.

4. Evaluating model performance using the Mean Absolute Percentage Error (MAPE) as the chosen evaluation metric. However, tasks with zero hours for certain days pose challenges to accurate MAPE calculation.

5. I'm considering applying stationarity tests to comprehend the data more deeply. However, I'm seeking guidance on effectively scaling this approach to encompass the complexity of the 8000-time series.

Despite these strategies, the challenge of scalability remains. For instance, training six models (ARIMA, ETS, BATS, TBATS, Prophet, and XGBoost) for each of the 8000 time series equates to a staggering 48,000 iterations without considering hyperparameter tuning. Given the substantial computation requirements, the practical feasibility of tracking these experiments using open-source MLOps tools like MLFlow is also a concern.

I'm reaching out for valuable insights and guidance. I'm keen to learn about best practices, practical approaches, and any available resources that can help me navigate this challenging endeavour.

Thank you in advance for your expertise and assistance.

• Time series are usually regularly sampled, e.g., on daily or monthly granularity. "Required work hours to complete various tasks" does not really sound like this. Do you really have a single observation (work hours) per time bucket? I suspect you rather have a large dataset with various attributes, and one attribute is a timestamp (which might be the time point where the task was started, or when it ended). Can you please clarify? Aug 6, 2023 at 11:06
• No, the dataset we have received does not include task start and end times. Nonetheless, the supplier, who provided us with the dataset, does possess this information for determining the time users take to complete tasks. The dataset I am working with consists of daily samples, where each team carries out a set of tasks on a daily basis. The ultimate aim is to forecast the projected work hours a team will need to complete that task in the next x months, based on their current task activities, even though precise task start and end times are absent. Aug 6, 2023 at 11:19
• Thanks. So you have multiple (or no) observations per task $\times$ team $\times$ day? I think it would be helpful if you could add a snapshot of your data into your question. I would assume that teams switch tasks now and then, and that team composition is not necessarily stable? Aug 6, 2023 at 11:23
• You might also be interested in the shortcomings of the MAPE. Aug 6, 2023 at 11:27
• don't use an ensemble of models. this is fine for kaggle, but impractical ( as you have realised) for normal professional use cases. just select one model and optimise that. Aug 6, 2023 at 23:03

First off, do not look to standard time series forecasting algorithms. These presuppose exactly one observations per time bucket, e.g., per day, week or month (and this observation may be "zero"). What you have, in contrast, is zero or multiple observations per time bucket. In addition, standard forecasting methods expect the time series to be continuous, but you may well have "holes" in the series where some teams do not work on some tasks or task types.

Instead, I would use standard regression models, "regression" being in the Machine Learning sense: predicting a numerical output. Just feed in your predictors and build models as usual.

If you suspect time dynamics, model these. Maybe your teams are less productive on Fridays? Then feed in Boolean dummies for day of week. Perhaps they are less productive during summer? Feed in a Fourier transform of the day of year. Possibly start with a multiple linear regression as a benchmark before trying more complex methods.

Think about what all those zeros in Hours are: did people really finish a task in zero time, or is that really a missing piece of information, or was the task open and they did not work on it that day? As always, understanding your data is usually much more important than tweaking models. You may want to look at zero-inflated models.

In your question, you show TeamID and TaskID. I hope you actually have task (and/or team) features so you can actually predict something, because TaskID sounds like an ID that was used for one task and will therefore not be used again - so you would not be able to forecast for a new TaskID. But again, this is standard ML.

Finally, the MAPE has major shortcomings, especially if we have zeros, whose treatments makes quite a difference. Either use it as an objective measure, or if you use a "standard" loss function like MSE or likelihood, you may want to post-process predictions to find the point prediction that minimizes the expected MAPE. Actually, I have never seen a business problem that was better solved using a MAPE-optimal forecast rather than an MSE-optimal forecast.

• Thank you, To clarify, when you suggest using standard regression models, are you recommending training a regression model directly on the entire dataset without considering the TeamID and TaskID? In other words, should I disregard TeamID and TaskID and build a regression model solely based on the predictor variables? Or do multiple time series forecasting considering Task and Team ID. Aug 8, 2023 at 10:49
• No, I would recommend training a model on all predictors that make sense, and TeamID is definitely one of those (since you want to predict future tasks for a Team you presumably have seen in your training data). It might make sense to instead use features of your teams, like number of developers, number of architects etc. in your team. And the exact same holds for TaskID: if you want to predict for a task you have already seen in your training data, use this as a predictor. I am just wondering whether this makes sense, or whether you should rather use task features. Aug 8, 2023 at 11:02