Friday, 9 December 2016

Week 6 [12.12-18.12.2016] Using Machine Learning to Predict Project Effort: Empirical Case Studies in Data-Starved Domains

Hello everybody.

I would like to introduce a Using Machine Learning to Predict Project Effort topic for you.
This study is trying to approach problems related to project estimation effort and how it can be solved with usage of machine learning algorithms.Ideally, software engineering should be able to use machine learning to control or significantly decrease the costs associated with building software. In reality, there are very few examples of applying such applications early in the software life cycle. One reason for the scarcity of examples is the lack of empirical data in the software engineering discipline. This dilemma is quite evident when constructing models to predict project effort.

  1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?
  2. Are you familiar with any existing solutions for this topic? What is state of the art in this field
  3. Do You know any open source data sets related to this topic?

35 comments:

  1. 1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    As it is written in article adopting a bottom-up approach for estimating project effort is quite feasible. And the accuracy variates between 90 and 100 percent. This approach offers very good potential, but I base only on article.


    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field

    Unfortunately I am not familiar with any existing solutions for this topic.

    3. Do You know any open source data sets related to this topic?

    Unfortunately I don’t know any open source data sets related to this topic. But I guess there are some neural networks based on open source data sets, maybe during artificial intelligence classes it was mentioned but I don’t remember at the moment.

    ReplyDelete
    Replies
    1. Thank You for comment. I understand that it is not easy to find data sets for proper model training.

      Delete
  2. 1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    In my view it is possible to apply such solutions for effort estimation for sure. This approach offers savings not only in money matters. It would help to estimate any project effort closer to the reality. Error of circa 13 percent is really petty comparing it with existing one of above 40 percent. For me it looks promising.

    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field

    Unfortunately I don’t have much in common with software engineering. All I know is from the article: just few cost estimation frameworks that were mentioned especially those using tuned COCOMO – software cost estimation model.

    3. Do You know any open source data sets related to this topic?

    It is really not my cup of tea however I’ve heard that Amazon and GitHub offer such data sets. On the other hand I don’t know if these sets are related to machine learning. As it is indicated in the article - learning when data is scarce is a very common problem – so I think that those data sets are different type.

    ReplyDelete
    Replies
    1. Thank You for Your opinion. I think predicting estimates can not be treated as holistic approach but can gives us great insight about organization delivery capabilities.

      Delete
  3. 1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    If humans can do it AI can do it as well. We just need to understand what it takes to design such AI to make it being capable of doing such a task.

    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field.

    I'm not really into the field I would have to google it to give you quick answer so I don't think it's the point.

    3. Do You know any open source data sets related to this topic?
    I think git repository or any big software repository has a lot of META data related to tasks and their progress of completion so I think it would be possible to extract those data and use it.

    Thank for great paper.

    ReplyDelete
    Replies
    1. Thanks for your comment. Github or Bitbucket could be truly great data sources for such research. As well Google lately open sourced large date with Stackoverflow questions and answers.

      Delete
  4. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    I think it is possible to apply machine learning or AI solutions for effort estimation. Machine learning is growing really fast so does its applications. Tho most difficult task in this topic is to collect proper data and data preprocessing. If it is done well I think that this solution could have significant outcomes.\

    Are you familiar with any existing solutions for this topic? What is state of the art in this field?

    I am not familiar with effort estimation but I am interested in machine learning and artificial intelligence. I think that some deep learning solutions may find application in this field, and it can be improved by some other ideas.\

    Do You know any open source data sets related to this topic?

    Unfortunately, I don't know any open source data set that could be related to this topic. Maybe someone who works with the topic o effort estimation can tell something more about it.

    ReplyDelete
  5. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?
    I would like to see something like that. We can't estimate effort properly as a human beings. How wrong AI will be in such situation?
    Are you familiar with any existing solutions for this topic? What is state of the art in this field?
    Planning poker is my favorite :) COCOMO, ANGEL to name few of them. I'm not the biggest fan of effort estimation, so it's hard for me to elaborate more about the state of the art.
    Do You know any open source data sets related to this topic?
    I'm afraid not. You can always check Stanford's data sets. They have plenty of different data from variety of sites.

    ReplyDelete
    Replies
    1. Planning poker is great. Thank You for recommending Stanford's dataset.

      Delete
  6. 1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    I believe that would be possible. We already have techniques to cope with large unstructured data generally known as Text Mining or in general Natural Language Processing.

    The only problem is, whether the management would accept any realistic estimates :)
    I honestly doubt that it will happen.

    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field?

    This area is outside of my personal interest (well, partially it is, but not the business side of the story), therefore I am not aware of any automatic estimation techniques, especially the ones employing Machine Learning.

    3. Do You know any open source data sets related to this topic?

    I haven't come across any of such dataset. And I wouldn't be too optimistic. Most of statistical inference problems are not about algorithms, but about data. It's doubtful that somebody would ever release something like this, it would be like killing the golden goose.

    ReplyDelete
    Replies
    1. Thank Paweł for fours comment. Maybe in agile environment management would not have big issue with accepting estimates from some kind of machine learning based solution. They are already accepting estimates from engineers any way.

      Delete
  7. 1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    According to the information found in the mentioned article adopting a bottom-up approach for estimating project effort is quite feasible. We can expect between 90 and 100 percent accuracy. It looks promising (assumption based only on that article)

    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field

    Unfortunately not.

    3. Do You know any open source data sets related to this topic?

    Unfortunately, I don't know any open source data set that could be related to this topic.

    ReplyDelete
  8. Yes, It looks promising. I think it can gives us another perspective how to estimate projects.

    ReplyDelete
  9. 1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?
    According to this article the approach for this problem is quite promising. Satisfying results of experiments between 90 and 100 percent are a good sign for the future.

    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field.
    I am not in this filed of research but i think Google, Microsoft and Facebook are working on machine learning technologies.

    3. Do You know any open source data sets related to this topic?
    I'm afraid i do not know any open source data sets for this filed.

    ReplyDelete
  10. Hello, thank you for sharing this article with us. Let me get straight to the questions:
    1. It's hard for me to predict if the outcomes would be of any significance, however I believe that AI can do a pretty solid job in almost any field, so why not. ;)
    2. Unfortunately I am not up to date with this topic.
    3. No, not really.

    ReplyDelete
  11. Hi, very good, interesting and important topic, thanks for sharing

    1.Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    Yes I think it is possible and moreover, I think it needs to be done. But I can see a drawback in devotion to the research. As mentioned in the introduction of the article, there is a lack of data in the software engineering discipline. And as companies may share these kind of information with scientists for the research purpose, I don't think companies will share it between them to achieve a better estimation results, as I consider this kind of information as a very valuable bussiness secret. But the topic is interesting and vary much can be done in this area.

    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field?

    Your post is the first thing that showed me that there are researchers who are working to solve the problem. So, I'm afraid I'm not familiar with any existing solutions.

    3. Do You know any open source data sets related to this topic?
    As clarified in my previous answer, this is a new area for me. So, unfortunately I don't know any open source data sets related to the topic. But I know a good Machine Learning Repository as I used it for some of my researches:
    https://archive.ics.uci.edu/ml/datasets/Open+University+Learning+Analytics+dataset

    Maybe you'll find something there

    ReplyDelete
    Replies
    1. Thank You for your comment. Maybe in the near future companies will work together on such topics or at least will create some kind of consortium. Microsoft has made some good example by open sourcing nearly 100K questions and answers in WikiQA dataset.

      Delete
  12. 1. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    I am certain that the idea of utilizing machine learning or AI solutions for any field is worth considering and to my mind I think that this is the future of majority of fields and disciplines. Effort estimation is among them as well. Traditionally the estimation of time required to deliver the project was based on historical data. First step was to identify the similar project and based on the hours spent on them, one could budget the effort/cost in the plan for their project. Main constraints of this approach are:
    A. team members of the project can be different therefore the assumption of similar hours spent on the project are biased in this manner.
    B. Projects are not identical - there would be potentially several bottleneck times during the project which project manager is unable to predict using historical data, however I think that this factor could potentially be overcome through AI.

    2. Are you familiar with any existing solutions for this topic? What is state of the art in this field.

    As far as I know company called "Open Mind" is working on similar subject. In terms of state of art in this field - I do not feel comfortable, to competently advise.

    3. Do You know any open source data sets related to this topic?
    Please endeavour searching through GitHub.

    ReplyDelete
    Replies
    1. Thank for pointing Open Mind project it seems to be really interesting research in this field.

      Delete
  13. I don’t think that we are in the moment when we can replace elements like effort estimation with AI and/or machine learning. Of course there are some task that anybody (or anything) can make a proper estimation but such tasks are only a small part of every project. In a real case scenario there is too many hard to describe elements that can have a significant impact on the estimation like for an example, way of communication between PM and the client. It is hard to measure it in scale from 1 to 10 but an experienced PM can make a proper estimation based on his/her experience.
    Maybe I’m an ignorant because I don’t track such solutions but based on what I read I’m sure that machine learning and AI are just not there yet. We have many very good ideas how to use those two concepts, but we still can’t to do that properly.

    ReplyDelete
    Replies
    1. Thnaks for your comment. In agile environment there is no such role as PM and management usually is accepting estimates from engineers.

      Delete
  14. Hello,

    Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?
    I think that it is possible to apply machine learning or AI solutions for fields like effort estimation. Moreover, it has been mentioned in the article the bottom-up approach can be used in order to explore effort estimation. It provides feasible outcomes producing 90-100% accuracy.

    Are you familiar with any existing solutions for this topic? What is state of the art in this field?
    I am afraid I don't have much to say about any other existing solutions for this topic. As far as I know Google is commited to such activites. One of the patents was reported 10 years ago https://www.google.com/patents/US8375364 .
    Do You know any open source data sets related to this topic?
    Unfortunately, I don't know any. Probably there are some provided via GitHub or maybe Google.

    ReplyDelete
  15. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?
    From my knowledge machine learning can be applied to every field from cleaning houses to surgeries so I think that you can apply ML techniques to this field. I don't if itwoulda have any significant outcome that is because it is not my field of interest.

    Are you familiar with any existing solutions for this topic? What is state of the art in this field
    No sorry I am not familiar with any existing solutions in this area.

    Do You know any open source data sets related to this topic?
    If there is no data set on Kaggle so it does not exists :D.

    ReplyDelete
    Replies
    1. Yes I know , unfortunately I didn't find any keyword on Kaggle. But still trying :)

      Delete
  16. Hello,
    thanks for the interesting topic. This article has interesting conclusion where we can find answer for your question. I want to say that I agree with thesis: estimating project effort is quite feasible :). I am not sure are we able to estimate project effort properly but it is possible for sure ;). If we ask few Project Managers about project we can receive couple different answers. Never the less it depends from factors that we will include in our calculations.

    I am not sure are we able measure an effort and I do not know any existing solution. Have you heard about any commercial or opensource tool that can measure an effort?

    ReplyDelete
  17. Thank you for sharing this artile with us.
    1. I am not good at predicting anything, but one thing I can be sure of, the results of experiments between 90 and 100 per cent, as per the article, seem rather promising.
    2. Unfortunately, it's the first time I encountered this topic, so I am not familiary with any existing solutions when it comes to this topic.
    3. Not really. But you inspired me to do some digging on the Internet and read more with regards to this subject. Thank you!

    ReplyDelete
  18. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?

    The conclusion of presented article suggest that it is possible to apply machine learning/AI solutions to estimate effort. It is a nice perspective, because from my experience it is a difficult task to do it in a IT project and quite often happens that estimations are not correct.

    Are you familiar with any existing solutions for this topic? What is state of the art in this field

    I only used PlanITpoker tool. Don't know any other solution.

    Do You know any open source data sets related to this topic?

    Sorry, but no. Actually I am not much interested in project management and effort estimation, so I never delved into the subject.

    ReplyDelete
  19. Do You think it is possible to apply machine learning or AI solutions for dry field as effort estimation and if would have any Significant outcomes?
    I think that this is possible, the article shows the way that we can follow.
    Are you familiar with any isnt solutions for this topic? What is state of the art in this field. Unfortunately, I do not know very well this topic.
    Do You know any open source data sets related to this topic?
    unfortunately I do not know

    ReplyDelete
  20. I don’t think the question of effort estimation should be taken into consideration at all. I’m not interested In the outcomes related to this estimation. It seems to be useless. I am afraid I’m not familiar with this topic from question 2. I wish I knew but I must admit I can’t discuss this issue because I don’t know much on that subject.

    ReplyDelete
  21. Article very interesting but unfortunately this is not my field - I thought it is certainly possible to estimate effort from the AL. Sorry I can not give concrete solutions in this area or dedicated software for data analysis to help in estimating the costs.

    ReplyDelete
  22. Do You think it is possible to apply machine learning or AI solutions for such field as effort estimation and if would have any significant outcomes?
    From my knowledge machine learning can be applied to every field from cleaning houses to surgeries so I think that you can apply ML techniques to this field. I don't if itwoulda have any significant outcome that is because it is not my field of interest.
    บาคาร่า online
    gclub online

    ReplyDelete
  23. I liked the arrows for their help in organizing your notes.
    doctor strange (2016)

    ReplyDelete