In this Data Science in Minutes, we will describe what time series forecasting is, and provide several examples of when you can use time series for your data. Time Series is looking at data over time to forecast or predict what will happen in the next time period, based on patterns or re-occurring trends of previous time periods.
Welcome to this short introduction to Time Series.
Whenever data or observations or something are recorded at regular time intervals,
you’re looking at Time Series data.
Time Series is looking at data over time to forecast or predict what will happen in the
next time period, based on patterns or re-occurring trends from previous time periods.
History often repeats itself, so whatever events happened in the past, they are likely
to happen again in the future.
The most common, basic example of a time series is seasonal sales revenue.
Each year during the holidays, sales revenue goes up, and during off seasons, sales go down.
This is not hard to predict, as we can almost expect what is to come every holiday season.
But you can also get other time-based trends in your data such as a consistent upward trend
or improvement in a company’s performance over time, or a downward trend, where the
company is consistently falling each year.. or month.. or day — whichever time period
What makes this different to other types of predictive models, is that the prediction
is based on a given time, looking at a sequence of observations over time.
We could look at a trend on a graph and take an educated guess that the trend will probably continue.
But a Time Series forecast can give us a better estimated figure for how much it could continue.
For example, given this time of day, a Time Series model predicts 100,000 people to login online.
You might have known there would be a lot of people online, but now you can plan for
how many additional servers and infrastructure you need for your online platform,
based on how many online users are predicted.
And you’d only use those servers for that particular time of day, switching them off
for the rest of the day when you don’t need them to save money.
Or, it could be that the model predicts a million online users in the year ahead,
significantly increasing from last year and even more so the year before then.
When you reach a point of continuing significant growth, you might decide now is the right
time to invest in better infrastructure for the year ahead and coming years ahead.
Another example is say you have a sensor device recording the number of vehicles that cross
an intersection every 20 minutes.
You could use these counts of vehicles to predict that in the next 20 minutes, traffic
at the intersection is likely to spike to a huge amount.
So now maybe your trip planning app could re-route your drivers to avoid this congested,
problematic intersection, distributing the traffic load more evenly across roads.
It is also possible to model data with no recognizable pattern or trend in Time Series.
When there is a trend or pattern, you can look back at the whole of history of your
data and see that pattern occurred over time.
But if there is no recognizable pattern, then your best bet is to rely more on what’s
recently happened and less on what’s happened far in history.
What’s happened recently is more useful in guiding us to what will happen next
than to look back at the whole of history, which only shows us pretty much anything could happen.
And that sums up Time Series in no time.
If you found this video useful, gives us a like.
Or, you can check out our other videos at tutorials.datasciencedojo.com
Learn how to setup a time series model:
Time Series in Python Part 1: Read and Transform Your Data
One Versus One vs. One Versus All
More Data Science in Minutes:
Natural Language Processing
More Data Science Material:
[Video] Time Series in Python Part 2: ARIMA Modeling and Forecasting
[Video] Time Series in Python Part 3: MAE Forecast evaluation
[Blog] Time Series Business Applications