Trend Analysis and Forecasts

Hi Marcio,

I’m not familiar with P3e. However I did a lot of progress S-curve in Excel.

Before Primavera P3, with derive the Progress S-curve using Gant Chart and eventually PERT/CPM. The simplest i can discuss to you is the Gantt Chart.

For each activity, you evaluate the duration, start date and finish date. For simplicity, we will use the early start dates in your Gantt Chart.

Each activity you assign a percentage weight. You can use your cost for each activities and compute the equivalent percentage. This equivalent weight for each activity will be distributed in the particular timeline of each activity. The simplest you can do is to distribute it uniformly, however this is not very precise. You can also distribute the percentage base on your understanding on the cost behaviour of each activities that you derive from past experience.

This is only my humble contribution. I hope it will help you jumpstart your S-curve

Best regards,

Charlie

Charleston-Joseph Orbe;

Thanks a lot for your response.

But i also need to create a trend curve. This trend curve will show me which will be the progress based in the progress until the moment.

If P3e can’t do it, i’d like to know how to do it with another tool (like excel).

Best regards;

Marcio Eduardo
Rio de Janeiro - Brazil.

first export the data to excel,then
prepare the date in excel and generate the s-curve,
then right click the s-curve, from the menu select ’add trendline, select the type of trendline most fit your s-curve.

hope this helps

Dear Zhang Haixiang;

Thanks a lot for your response.

It realy works on Excel. But do u know how to create trend curve only from the last update? As a continuation of the progress curve.

Best regards.

Marcio Eduardo.
Rio de Janeiro - Brazil.

you have x-axis the date, and you have the y-axis the progress. the data may like this (current date is 6):

date 1 2 3 4 5 6 7 8 9
progress x1 x2 x3 x4 x5 x6

select all the above data for you curve.
the s-curve will stop at date 6 (which is you last update),but the trend line will extend to date 9 (last data of your x-axis)

to make it start only from you last update and look like continuation of your s-curve, my suggestion is to do it manually.

add anothor row of date for your trend-curve, try to put some data, and make it fit with the trendline, then delete the trendline

hope this helps

Marcio,
Earned value analysis forecasts EAC basing on the trends (performance factor). Performance factor is usually set as 1/CPI. Obtaining report on future costs and multiplying them by performance factor you will adjust them in accordance to trends based on the past performence.
In response to your question about other tools - Spider Project creates trend curves for the data you choose.
Best Regards,
Vladimir

Zhang Haixiang and Vladimir Liberzon,

Thanks a lot for your responses. I’ll try both.

Do u use to create trend curve to analyse the future and create recover plans???

Best regards.

Marcio Eduardo
Rio de Janeiro - Brazil.

Marcio,
we insist on management by trends. Not only S-curves but trends of other parameters like planned finish date, resource requirements, probabilities to meet target parameters successfully shall be analyzed. If the trends are negative then it is a trigger for considering corrective actions.
Even if current project status is favourable you shall analyze trends.
Example: you are ahead of schedule for 2 days, a week ago you were ahead of schedule for 5 days, a month ago - for 10 days. This trend means that you have a problem that shall be solved immediately not waiting until the project will be in trouble.
Regards,
Vladimir

Marcio,

I’m used to calculating it manually based on performance factor. It is calculated as Actual / Planned. Normally latest figure is used although you may be requested by client to use average figure which in case of poor previous performance and a sudden increase will not allow you to show any significant improvement until you performance becomes stable.

Alex

Thank u all for your responses.

Good explanations here. It is realy going to help me.

Best regards;

Marcio Eduardo
Rio de Janeiro - Brazil.

I feel like you’ve explored only one side of the trend analysis: the "statistical" side using only the numbers. But you need also to check a more reality anchored analysis.

Part of the trend analysis shall also rely on reveiwing the remaining durations and future durations, in-light of past performances, asses whether the past perf is relevant or not. Once the reviw is done update the plan accordingly and get out your forecast curve. Now you can compare this to the "statistical" trends for an even better analysis.