Cost Estimating – Part 3 of 3

On this last episode of the series on cost estimation, Ricardo talks about the proportionality factor, parametric modelling and neural networks.

Download (11 MB)

Episode transcript

The transcript is generated automatically by Podscribe, Sonix, Otter and other electronic transcription services.

Hello, everyone. Welcome to the five Minutes PM podcast today. I like to finish the series of Cost estimation. I want to finish by teaching to you the two final waste of Estimating Cost. The first one is the proportionality back door in the proportionality factor, for sure. It's also known by Lang and hand factor. Lang developed this in the forties to forecast the cost of factory, base it on key elements or key equipment or key cost of that factor.

And then based on this, study's he try to find factors that you can move to apply this major cost items to forecast the final cost. So let me give you a couple of examples here. First let's sink on an oil refinery and the oil refinery is very much dependent off the main equipment's that distillation tower and the order major equipments. So let's suppose that this major equipments will cost $2 billion.

Then I have the Lang factor for fluid processing, and this is a known number that is four point 74. There is a Lang table to help you to understand this. So if you moved to by 2 billion times, for point 74, you have 9.5 billion, and this will help you to estimate. So let's suppose now, of course, your listen to me and say, Oh, but Ricardo, I don't work in the oil or a petrochemical. That's no problem. You can use the same concept doing an analogy.

Let's suppose you have a restaurant and let's suppose that you're main cost item is the stove. Then you say, okay, own this past temp projects. I know that I spent a thousand dollars to buy the OVN and the total cost of the construction was 10,000. Then another project I bought 2000 stoves and then the final cost was 22. Then I start dividing the total cost by the cost of the main item demo, see a 10%, a 10.1, nine point 87.

Then I calculate the average. And then I can the best my new restaurant by saying, look, you take the cost of your stove, then your move to apply by the proportionality factor of 10. And that, you know, you go to the market in C. Now the cost of one stove is 1,500. Dan, the cost of the restaurant will be 10 times that because now you know that the factory Stan. So it's very, very interesting way, but there's will only work if you have major components and you need to understand also that this has an acceptable range of minus 30% plus 50%.

This is the NC standard. So you need the understand that it's not true. You can not please build a refinery. Think in that you take the distillation tower and multiply by a factory. It's a very simplistic, but the factor will help you to understand the future. Cost the last but not least is the parametric modeling. And this is, it's not so easy to explain because this, I need to model to create the model based on project data based on physical or functional characteristics.

So what do I do? I have project data. I have components of this data, and then I can create an equation that can be a linear or a new linear equation. So I try to understand, okay, the cost of when to keep month based on the temporary or so let's suppose a refrigerator, the cost of the refrigerator is proportional to the reduction of temperature. So you can start creating this kind of model. Look, if your listen to what I'm saying, you think this is quite complex and yes, it's very hard for you to create this formless.

And this is why I wanted to teach you something that blew my mind. When I started doing is the concept of neuron network. What is a neural network? It's a set of learning algorithms, very commonly Plex. And this is now on the only available because there is software that does that for you. For example, I use a softer that I like very much. It's a part of the at risk chute from Palisade called neuro tools in what is the neuro to its a tube based on an XL that does this soft based on data that I provide.

It'll try to model in create plus to help me to understand. So let's give you a very quick and simple example, let's suppose that I'm a restaurant company and I have information about previous restaurants. I built an, the future and that they use based on the profile off the region. So I know the cost of the restaurant that I built. Okay, 10 restaurants. Then I know the region. I know that distribution of wealth on that region. So I know that 21% it's a gets last than X amount of money.

The 25, between the two values. Then I have some distribution between male men and women. I have this age breakdown. So I have some, the demographics about that restaurant and the Cost and the revenue. Okay. And I want to do a feasibility studies of a new one. So then I ask the software and that neural networks to find out, is there any correlation between profit, an age, profit and gender profit and location profit, and for example, cost of construction, then I run somebody like a simulation and then the software creation, it's a formula for me.

And then for my next restaurant, I just need to say, okay, we're going to do you want to place this restaurant? Do you have the demographic information of that region? You want to put the restaurant based on age, gender, net income, and then base on this data. The neuro network can count for me the cost, and this is a very complex process, but it's so effective. So if I can give you an advice, this is the way I love using Today are a force to model.

It is not so simple. Of course we would love to do just physical dimensions. And it's very okay, how big is the restaurant? And then you just multiply. But this is very, very simplistic. The use of neural networks can leverage your cost estimation heavily in the past. We didn't use that because it was just impossible because the easy for us to do that was just insane. And now it's quite simple to do so this is what I try to demonstrate to you and why I spend three Podcast to explain this because most of the people do very weak cost estimation, and they do using very simplistic models and then they fail.

So it's very, very important. I would love if you have five minutes now, just go to the web and put neural networks and study five minutes. You say that that this could help you to leverage your Cost Estimating much higher and not with a painful process, like doing a complete deterministic model of the project budgeting. So think about that. I hope you enjoy the series. I wish you very successful cost estimation and see you next week with another five minutes BM Podcast.

Menu

Resources & Insights

Products & Services

Ricardo Vargas

Podcasts

Cost Estimating – Part 3 of 3

Search in Podcasts

Latest podcast episodes

Book recommendations

Menu

Resources & Insights

Products & Services

Ricardo Vargas

Podcasts

Cost Estimating – Part 3 of 3

Search in Podcasts

Latest podcast episodes

Book recommendations

Subscribe to our newsletter