Getting Started with Orange 06: Making Predictions

แชร์
ฝัง
  • เผยแพร่เมื่อ 16 ธ.ค. 2024

ความคิดเห็น • 126

  • @gero8049
    @gero8049 5 ปีที่แล้ว +7

    Man, thats great. I am studying data science for 9 months. I wish I discovered this on the beginning.

    • @DJVARAO
      @DJVARAO 5 ปีที่แล้ว

      I feel you. I have been working on these things for many years and no tool have been this easy to use before.

  • @adversun
    @adversun 7 ปีที่แล้ว +13

    With the current data-set, I am getting all 3 fruits with Classification tree and with Logistic regression. I think it's because this video is demonstrated with 39 data-sets but the actual data-set is 35.

    • @hafizmaruf2654
      @hafizmaruf2654 5 ปีที่แล้ว +2

      same with me

    • @henchhh
      @henchhh 5 ปีที่แล้ว +1

      me too

    • @v1tr4elits
      @v1tr4elits 3 ปีที่แล้ว +1

      Yes same with me too, but i try to add some test data that taken from training data which result is vegetable and it predict right as it is

  • @milenageorgieva2233
    @milenageorgieva2233 7 ปีที่แล้ว +1

    The data in the test data set is different. Check the second row, column vitamin C %. It should be 9 as in the video. Second, if you check the Tree, you will see that the training data is completely different from the training set provided by the link above. So, that's why you can't get correct fruit-vegetable-fruit prediction.

  • @alisterrebello5337
    @alisterrebello5337 2 ปีที่แล้ว

    2:23 what does it mean by giving the widget some classification model?

  • @shaz-z506
    @shaz-z506 5 ปีที่แล้ว +5

    Thanks, but I want to know how I can handle situations where the class is an imbalance?

  • @migoz95
    @migoz95 2 ปีที่แล้ว

    Why do I get error: input contains nan infinity or a value too large for dtype('float64') in the prediction widget when trying my own dataset

  • @HashCat84
    @HashCat84 8 ปีที่แล้ว +19

    perfect thank you so much, saved my wife PHD thesis

    • @Mkom-ij6sm
      @Mkom-ij6sm 4 ปีที่แล้ว

      succes for your wife brtotheer

  • @pengenjadiilmuan9631
    @pengenjadiilmuan9631 4 ปีที่แล้ว +1

    Any widget with neural network for prediction?

  • @ganeshkamath89
    @ganeshkamath89 ปีที่แล้ว

    With Orange 3.35, the data from Tree is not predicting the second row as a vegetable, but logistic regression is still classifying it correctly. I think there is some problem in the Tree classifier in the latest version.

  • @rayzek4347
    @rayzek4347 2 ปีที่แล้ว

    the software does not show me the approximation to the reality of the prediction (the blue and red bar). :(

  • @eylmaz6696
    @eylmaz6696 10 หลายเดือนก่อน

    is this system uses Apriori algorithm ? (association rule ) ? in assosication rules ?

  • @steril87
    @steril87 7 ปีที่แล้ว +4

    I just can't find the classification tree in my orange widgets, is it some kind od optional addon?

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว +7

      No, it has been renamed to Tree and is now is Model pane.

    • @steril87
      @steril87 7 ปีที่แล้ว

      Oh thanks, that's why I got confused :)

  • @atassano2001
    @atassano2001 ปีที่แล้ว

    Great tool!! But where can I download all the datasets? The datasets provided with the instalation are only a few of them

    • @OrangeDataMining
      @OrangeDataMining  ปีที่แล้ว

      The Datasets widget.

    • @johannhmartinez8550
      @johannhmartinez8550 หลายเดือนก่อน

      @@OrangeDataMining There is no such a data set, or at least in my UBUNTU orange3 installation... What should I do? Thanks

    • @johannhmartinez8550
      @johannhmartinez8550 หลายเดือนก่อน

      @@OrangeDataMining Or at least would yu please tell us what is the name of this dataset? tks!

    • @OrangeDataMining
      @OrangeDataMining  หลายเดือนก่อน

      @@johannhmartinez8550 There's no name. We collected the data from Wikipedia, but did not store it for future use. Any classification dataset from Datasets will work, as described in the video description.

  • @mattmatt245
    @mattmatt245 4 ปีที่แล้ว +1

    What if based on ROC analysis, I want to use a different threshold (say 0.3) for my predictions ?

  • @jim3xPRO
    @jim3xPRO 8 ปีที่แล้ว +9

    I feel love.
    Thank you very much!

  • @danialali4133
    @danialali4133 5 ปีที่แล้ว +3

    Thank you for this amazing work >> please tell me why with the current data-set, I am getting all 3 fruits with Classification tree (and like your result with Logistic regression).

  • @Alex72RM
    @Alex72RM 6 หลายเดือนก่อน

    Hi! Hope you could update link to datasets that are actually broken.

    • @OrangeDataMining
      @OrangeDataMining  6 หลายเดือนก่อน

      Sorry, but the datasets are no longer available. You can follow the same procedure with the simple Iris dataset from the File widget. The new videos are available here: th-cam.com/video/f4QhPmWNzP0/w-d-xo.html

  • @adamgrant9085
    @adamgrant9085 8 ปีที่แล้ว +4

    Great video, it really helped me out with analyzing my own data! Would I be able to get a copy of the fruit and vegetable dataset by any chance?

    • @OrangeDataMining
      @OrangeDataMining  ปีที่แล้ว +1

      Ugh, this is a very old dataset and we messed it up somewhere along the way. I suggest you have a look at our refreshed videos with new datasets for classification.

  • @donm99979
    @donm99979 10 หลายเดือนก่อน

    Hi, can you upload again and share the file that used in this video? The link is '404 not found'

  • @MarioDeCeglie
    @MarioDeCeglie ปีที่แล้ว

    I found other examples in the book "Data Science and Engineering - A learning path - Volume 2 Exploratory Data Analysis, Metrics, Models: with applications in the Orange Python-based environment". It could be useful to know.

  • @geetanjalisahi4512
    @geetanjalisahi4512 ปีที่แล้ว

    the dataset is not available even through the link, please help

  • @landongray1354
    @landongray1354 8 ปีที่แล้ว +1

    tl;dr The dataset in the video and the description are not the same
    If you go to 0:51 and pause to look at the table. On the left panel under info it says 39 instances while in the actual data set there are 35 instances

    • @OrangeDataMining
      @OrangeDataMining  8 ปีที่แล้ว +2

      They are indeed different. The shorter one is missing cabbage, kale, lime and cranberry. But I don't think it's anything major. The principle remains the same.

  • @TheIdea4life
    @TheIdea4life 7 หลายเดือนก่อน

    test dataset url is not working either

  • @kushalsharan9622
    @kushalsharan9622 2 ปีที่แล้ว

    While joining Linear Regression and Test data to prediction, it says 'Yield not in domain' where yield is my target variable.

  • @Richard-pp9jr
    @Richard-pp9jr 3 ปีที่แล้ว

    The rank widget needs to save its state, and load state. I use rank first and train models on the best variables each time I save it the wrong click the selections are lost and investigation has to happen all over again. Can you post me a link to so I can be a feature creep thanks.

  • @mariavv3986
    @mariavv3986 ปีที่แล้ว

    where can we get the fruit and vegetable data¡

  • @TSUKILORD
    @TSUKILORD 7 ปีที่แล้ว +1

    Hello community! Hello Orange Data Mining team. Could you please let me know what exactly is the Decision Tree algorithm that is used for the experiments? ID3? C4.5? Or which exactly? It is kinda confusing since the widget, allows actually to tune it using different criteria and I am not sure if the default version is ID3 or C4.5 or what :S - thank you!

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว +3

      Neither. We wrote our own trees to make them fast, reliable and able to handle both continuous and discrete variables properly.

  • @romank1280
    @romank1280 2 ปีที่แล้ว

    In Orange 3.32.0 Tree model returns 3 fruits for me, while Logistic Regression is same as yours.

  • @migoz95
    @migoz95 2 ปีที่แล้ว

    Does orange data mining support multi target predections ?

  • @sanskarkhandelwal
    @sanskarkhandelwal ปีที่แล้ว

    Explanation is crystal clear... well made

  • @samarabbadi2523
    @samarabbadi2523 6 ปีที่แล้ว +3

    thank you for this video but i didn't get the same result using Classification tree and logistic regression

  • @SajalKantiGhosh
    @SajalKantiGhosh 10 หลายเดือนก่อน

    Hi, I can't find the dataset.
    Would you please help?

    • @SajalKantiGhosh
      @SajalKantiGhosh 9 หลายเดือนก่อน

      @Biohazard262 Not yet, unfortunately 😢

  • @namanshah1003
    @namanshah1003 7 ปีที่แล้ว

    Hello I am following the video exactly but I am getting an error "One or more predictors failed" I have 2014 data and I am trying to predict 2015 data. I have the same number of features in both data sets. Any ideas?

  • @subramanianramajayam2467
    @subramanianramajayam2467 4 ปีที่แล้ว

    how do i set the workflow to predict the unknown target values ?

  • @annamathew9110
    @annamathew9110 3 ปีที่แล้ว

    Is there any vedio that explains about the PLS widget .

  • @meenu5296
    @meenu5296 11 หลายเดือนก่อน

    Not able to download the datasets
    please provide the link again

    • @OrangeDataMining
      @OrangeDataMining  11 หลายเดือนก่อน

      Please see the new series for videos on predictions. This is one is deprecated.

  • @Sieg_W
    @Sieg_W 3 ปีที่แล้ว

    Hi and sorry for the obvious question. When I open orange I can't find 'Classify' and 'Regression' on left menu. Any idea please? Do I need to install them separately? Thanks

  • @bobking7454
    @bobking7454 7 ปีที่แล้ว +1

    There is no classification tree in orange 3.4.5 (the latest version), anyone can help?

  • @9999saisandeep
    @9999saisandeep 7 ปีที่แล้ว

    I was trying the same steps and my model(Logistic and Classification) is predicting all the 3 test observations as Fruit. I am using Orange 3.8. Am I doing something wrong.

  • @imtiazhossain1579
    @imtiazhossain1579 ปีที่แล้ว

    Can You please provide the excel file for prediction?

  • @bbmtau
    @bbmtau 5 ปีที่แล้ว

    Same result for logistic regression and SVM but all fruits with classification tree. Still wondering how would you tell which one is a better model. The output doesn't seem to say much.

    • @OrangeDataMining
      @OrangeDataMining  5 ปีที่แล้ว +1

      This is why Test & Score reports scores. Normally, you'd look at AUC. The higher the better. The model, which gets the highest AUC with cross-validation, is usually the best model. You can, of course, use a separate test data and check how it performs there.

    • @bbmtau
      @bbmtau 5 ปีที่แล้ว

      @@OrangeDataMining thanks a lot, I noticed you explained that very well in the next video.

  • @TheIdea4life
    @TheIdea4life 7 หลายเดือนก่อน

    train data set url is not working

  • @ndevtuts1472
    @ndevtuts1472 5 ปีที่แล้ว

    I can't find the classification tree widget in orange software.so please help me.

  • @laurinhads27
    @laurinhads27 ปีที่แล้ว

    I must have my y and my x values at the same file?

  • @EnricoRossi85
    @EnricoRossi85 7 ปีที่แล้ว

    Hi, I found that training data set linked here is different by that you used in video. Using training data set linked here I find all fruit predictions (with decision tree) and fruit-vegetable-fruit predictions with logic regression...why? Thank you

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว

      Because trees tend to overfit and do not generalize well. You can drag Tree Viewer from Tree and observe the structure of the tree. You will see that the tree built is the same as the results you are getting with predictions.

    • @jeandalay7630
      @jeandalay7630 ปีที่แล้ว

      @@OrangeDataMining does that mean that the logistic regression have a better prediction than the tree?

  • @knalli7129
    @knalli7129 9 หลายเดือนก่อน

    I have the same issue with the dataset...

  • @robevans2114
    @robevans2114 4 ปีที่แล้ว

    I do not see "Classification Tree" model (O 3.26)??

    • @bhargavig5257
      @bhargavig5257 4 ปีที่แล้ว +1

      It has now been renamed as Tree

  • @tirgdigital
    @tirgdigital ปีที่แล้ว

    It didn't work on my version 3.35.0.

  • @vimalmangeshkar9486
    @vimalmangeshkar9486 6 ปีที่แล้ว

    Hi...can you please tell me how do we split a data set, say in a 80-20 proportion for training and test data, in orange?

  • @andrewmccown3841
    @andrewmccown3841 4 ปีที่แล้ว

    Is there a way to view the specifications of the model after it is created? (i.e., the coefficients)

    • @OrangeDataMining
      @OrangeDataMining  4 ปีที่แล้ว

      Yes, you can inspect some models (LogReg & NaiveBayes with Nomogram, LinReg with Data Table, SVM with Scatter Plot).

  • @cvt3641
    @cvt3641 5 ปีที่แล้ว

    Thank you for your support. I just wondering that How to convert the Data table result to excel or other kinds of output? Thanks

    • @OrangeDataMining
      @OrangeDataMining  5 ปีที่แล้ว +1

      Use Save Data widget. It supports Excel files, too.

    • @cvt3641
      @cvt3641 5 ปีที่แล้ว

      @@OrangeDataMining thank you for meaningfull reply me. Other question is how limited of input data sorf can deal with (I mean the number of collomn land the rows of input data?)

  • @inbalsolomon7924
    @inbalsolomon7924 7 ปีที่แล้ว

    Both my training and test datasets have the same number of features and my classification column is labeled as 'target' in both files. However, when I connect my trained data (in this case, SVM) to my Prediction, I get a dotted line and nothing happens. Help please :)

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว

      Could you please submit a screenshot?

    • @inbalsolomon7924
      @inbalsolomon7924 7 ปีที่แล้ว +1

      Here's the pic of my setup: imgur.com/kuHgbNC

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว

      Well, you didn't provide Predictions with any model, which was trained on the training data, so it cannot predict. See how the output of Preprocess is 'Preprocessor'? This outputs only the steps for preprocessing, not the preprocessed data. You would have to send the data to SVM and/or AdaBoost.
      Also, you should never have two preprocessing widgets for two separate files! This leads to overfitting. Orange is designed so it handles preprocessing internally. Once you provide it with preprocessed training data, it will apply the same preprocessor to the test data.
      Finally, when connecting in this way, you could provide SVM on its own as a Learner to AdaBoost, without passing data through it. This would require two inputs to AdaBoost: Data and Learner (SVM).

    • @inbalsolomon7924
      @inbalsolomon7924 7 ปีที่แล้ว +1

      Great, that's super helpful and has helped resolved my issue! Thank you.

  • @MrTkmmkt
    @MrTkmmkt 8 ปีที่แล้ว

    I got error "Data has no target variable" when trying to create Classification Tree with your training set. What should I do for fixing it? I'm using Orange 3.3.7. Thanks.

    • @MrTkmmkt
      @MrTkmmkt 8 ปีที่แล้ว

      I found the way for fixing this error. I saved your data file to local, manual added 02 rows (variable type and variable kind) as in "Getting Started with Orange 04".

    • @OrangeDataMining
      @OrangeDataMining  8 ปีที่แล้ว

      That's it, you need to specify the variable you're building the tree by, that is your target variable (or class). You can also do this directly in the File widget by double-clicking the feature you want to modify or in Select Columns.

    • @danielalmeida7126
      @danielalmeida7126 8 ปีที่แล้ว

      Just click on the file and where it says features, meta, etc, click on the row you want to change and change it to target. As Orange said, this is to base your tree on a variable. I also noticed that if you connect a file directly to the tree classifier, it will say 'discrete class variable expected'. To fix this, just add the 'discretize' function between the file and the decision tree function. So file -> discretize function -> decision tree function -> tree viewer. And voila! I ran a test tree this way with 19,700 data rows and 7 variables. For such a big number I recommend clicking classification tree and typing 100 next to 'do not split subsets smaller than'. This means that it won't look to micro identify data under 100, which is kind of pointless as you don't want to get too specific, as you will be unable to see patterns. For example, before I had it set to 2 and each level on the tree would add another level to specify even more, to the point where it diluted the data into groups of 2-10 people. That's hardly enough to come up with a good pattern.

  • @sanarasheed6385
    @sanarasheed6385 7 ปีที่แล้ว

    this dataset is not present in Orange dataset directory. Where I can find it? Any link please?

  • @yakmag7761
    @yakmag7761 7 ปีที่แล้ว

    Hi am also using the Orange chapter 6 when i'm click the predition widget open the diffrent image not a higlight the same video

  • @duyanhnguyen8579
    @duyanhnguyen8579 7 ปีที่แล้ว

    Hi,
    I've been following this tutorial quite carefully, but i got 'mismatching target (classification)' error. Classification column is the target variable in my dataset but there is no classification column in my test data, is that the problem?

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว +2

      This is bug in the latest release and we're are fixing it as we speak. We apologize for the inconvenience.

    • @duyanhnguyen8579
      @duyanhnguyen8579 7 ปีที่แล้ว +1

      Great. Thanks for informing me :) Hopefully it will be fixed soon!

    • @jesquiagola
      @jesquiagola 7 ปีที่แล้ว

      I have the same problem .. I can not predict the results

  • @biljanajovanovic365
    @biljanajovanovic365 7 ปีที่แล้ว +1

    What dataset is ths? thank you :)

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว +1

      It's the data set from the description.

  • @詹天貴-x9k
    @詹天貴-x9k ปีที่แล้ว

    Thank you, Orange team, for providing the videos for us to learn the operations. However, it seems that there are errors in the information you provided. It would be better if we could practice with data that matches the examples shown in the videos, so that we can verify the correct results.

    • @OrangeDataMining
      @OrangeDataMining  11 หลายเดือนก่อน

      Please see the new series for videos on predictions. This is one is deprecated.

  • @sarmadkhan766
    @sarmadkhan766 2 ปีที่แล้ว

    I'm getting different results on the test data set.

  • @kotai2003
    @kotai2003 5 ปีที่แล้ว +1

    Thanks. This courses help me a lot.

  • @emensonjean7424
    @emensonjean7424 5 ปีที่แล้ว

    can i use this same to predict an outcome of a basketball game

    • @DJVARAO
      @DJVARAO 5 ปีที่แล้ว +1

      I doubt it. The amount of data you need versus the complexity will give you poor predictions.

  • @rossdemtschyna1314
    @rossdemtschyna1314 7 ปีที่แล้ว

    Hello, i get an error above my 'predictions' widget 'red X' and the error is 'tree mismatching target classification'. What does this mean? I am into video 2min and 27 seconds. I love this Orange package it really makes data science and algorithms simple.

    • @OrangeDataMining
      @OrangeDataMining  7 ปีที่แล้ว

      It means you don't have the same class values in your train data and test data. Probably your classifier was trained on a different data set than you then try to predict on.

    • @rossdemtschyna1314
      @rossdemtschyna1314 7 ปีที่แล้ว +2

      Thank you very much - my data was different which surprised me as i used your 'Fruits and Veg' links above.
      I really like where Orange is going and the possibilities it is offering.

  • @abdulrab4411
    @abdulrab4411 5 ปีที่แล้ว

    How can I make future prediction let say i want to know future profit...How to read the predictions.....I have no data science experience

  • @zigmontmackonis6548
    @zigmontmackonis6548 6 ปีที่แล้ว +1

    I use SAS, but Orange looks good.

  • @CrazyHunk14
    @CrazyHunk14 6 ปีที่แล้ว

    thanks a lot it is very useful for my data mining assignment . u taught clustering very well can u also make some videos on data mining algorithms and methods it will be very helpful . ps. u look cute in glasses!!!!

  • @josebarbozagonzales6460
    @josebarbozagonzales6460 4 ปีที่แล้ว

    the tree say me, all are fruits why?

  • @sslahmzmvr6169
    @sslahmzmvr6169 2 ปีที่แล้ว

    Character In the video It's great, I like it a lot $$

  • @MuhammadUsman-fg8dg
    @MuhammadUsman-fg8dg 4 ปีที่แล้ว

    I have tried with the same data but the tree prediction and Logistic regression values are changed...

    • @HcDaN
      @HcDaN 3 ปีที่แล้ว

      yeah

  • @fsconrado
    @fsconrado 2 ปีที่แล้ว

    Meu irmão e eu levando horas pra programar!!! antes de conhecer esse programa!

  • @Acosta360
    @Acosta360 7 ปีที่แล้ว

    I'm not getting the same results and yes I'm using the same data sets. I'm using the latest Windows version i.imgur.com/P9lY41A.png

  • @miguelgoncalves6112
    @miguelgoncalves6112 5 ปีที่แล้ว +1

    muito lindo

  • @hirakmondal6174
    @hirakmondal6174 5 ปีที่แล้ว +5

    I came here for some naughty orange comments but found that programmers are tooo serious these days..! (sigh)

  • @bodom11716
    @bodom11716 4 ปีที่แล้ว

    no random forest dislkiek

  • @anahisuarez1537
    @anahisuarez1537 8 ปีที่แล้ว +1

    (y)

  • @skaterfreak7658
    @skaterfreak7658 ปีที่แล้ว

    the dataset link isn't working

  • @excelforestate
    @excelforestate 8 หลายเดือนก่อน +1

    the dataset is not available even through the link