---
updatedAt: 2025-02-11T23:35:44.000Z
agentTools:
  projectIndex: https://docs.lifesight.io/llms.txt
---

# How to select planning base period

Use seasonality, trends, and holidays to select the optimal forecasting base period

### Overview

Selecting the optimal base period for forecasting involves understanding the underlying patterns in your data, including `seasonality, trend, and holiday effects`. Graphs that depict these elements can guide you in choosing a base period that will improve the accuracy of your forecasts. (Below graph is an example of that)

![](https://files.readme.io/66afb4457cbceb8ae36525f1fabe7bddd13f4e756250fe6307a4d932dcdcb467-Screenshot_2024-08-27_at_11.12.18_PM.png)

<br />

### 1. Understanding Seasonality

**Seasonality** refers to periodic fluctuations in data that occur at regular intervals, such as weekly, monthly.

**Example :**

The second graph (above) illustrates the seasonal component of the data:

* Clear cyclical pattern that repeats annually
* Peaks around $400K to $600K
* Troughs around -$400K to -$600K
* This pattern suggests strong yearly seasonality in the revenue data.

  **In these cases, you can select the`past year same month` as the `optimize based on` period for forecasting.**

> 📘 **Note**: Our model can identify seasonality and trends in your data and provide you with a comprehensive chart.

<br />

### 2. Analyzing Trend

**Trend** represents the long-term direction in the data, indicating whether values are generally increasing or decreasing over time.

**Example :**

The third graph shows the overall trend of the revenue:

* A gradual, steady increase over time
* Starts around $700K at the beginning of the period
* Ends around $850K by the end of the period
* This indicates a slow but consistent growth in the underlying business

<br />

### 3. Identifying Holiday Effects

**Holiday Effects** are anomalies or deviations from normal patterns that occur around holidays or special events.

**Example :**

* The bottom graph represents the impact of holidays or special events:
* Mostly flat around $0, indicating no effect for much of the year
* Occasional sharp spikes, with two major ones reaching about $1.5M
* These spikes likely correspond to specific holidays or events that dramatically boost revenue

<br />

### Overall Recommendation:

If you are forecasting for a month or quarter :

1. **If the data is seasonal and there is a flat trend**: You take last year's same month as the base period.
2. **If the data is seasonal and there is an additive trend**: You can either take last year's same month as the base period or the latest month as the base period.
3. **If the data is not seasonal and there is an additive trend:** Take the latest month as the base period.
4. **If the data is not seasonal and there is a flat trend:** Take the latest month as the base period.

> 📘 Note
>
> * You can understand why the proper base period is required, here.
> * If you are not able to figure out the proper base period, please reach out to <support@lifesight.io> or the respective account manager assigned to you.