TileStats
TileStats
  • Видео 163
  • Просмотров 1 906 849
Stochastic gradient descent (SGD) vs mini-batch GD | iterations vs epochs - Explained
For more videos in a logical order, go to: www.tilestats.com
1. How gradient descent works
2. Batch gradient descent (06:09)
3. Stochastic gradient descent (13:25)
4. Mini-match gradient descent (16:25)
5. Epochs vs iterations (18:00)
Просмотров: 225

Видео

How to compute a p value and extract a critical value in R
Просмотров 387Месяц назад
www.tilestats.com 1. Compute a p-value from a t-distribution (2:14) 2. Extract a critical value from a t-distribution (04:36) 3. How to plot the t-distribution (07:09) 4. The standard normal distribution (08:30) 5. The f-distribution (09:10) 6. The chi-square distribution (09:58)
Multinomial logistic regression | softmax regression | explained
Просмотров 909Месяц назад
For more videos in a logical order, go to: www.tilestats.com 1. Binary logistic regression 2. One vs all logistic regression (03:00) 3. The softmax function (06:40) 4. Multinomial logistic regression (07:40) 5. Multinomial logistic regression in R and SPSS (11:40) 6. Multinomial logistic regression vs ANN (13:30)
Why we divide by n-1 when calculating the sample variance - the proof | unbiased estimator
Просмотров 1,3 тыс.2 месяца назад
www.tilestats.com 1. The sample variance (1:00) 2. Sample variance vs population variance (02:40) 3. Unbiased vs biased estimator (03:45) 4. The mathematical proof (05:30) 5. The expectation rules (12:30)
Expected value vs mean
Просмотров 7132 месяца назад
For more videos in a logical order, go to: www.tilestats.com 1. Expected value vs mean 2. Weighted average vs the expected value (04:36) 3. Expected value of a discrete uniform distribution (06:35) 4. Expected value of a continuous uniform distribution (08:28)
Bootstrap confidence intervals - explained
Просмотров 8722 месяца назад
See all my videos at: www.tilestats.com/ 1. How to calculate a 95% bootstrap confidence interval (01:40) 2. Do bootstrap confidence intervals work? (05:00) 3. How to calculate a 95% bootstrap confidence interval for a difference in means between two samples (10:36) 4. How to calculate a 95% bootstrap confidence interval for parameters in a linear regression model (11:36)
How to identify and deal with outliers | The 1.5 IQR rule | Boxplots
Просмотров 5513 месяца назад
See all my videos at: www.tilestats.com/ 1. How to calculate the IQR (0:26) 2. The boxplot (02:45) 3. The 1.5 IQR rule (04:00) 4. How to deal with outliers (05:00) 5. When a boxplot does not work (06:54)
Bayesian statistics - the basics
Просмотров 2,4 тыс.5 месяцев назад
www.tilestats.com/ 1. t-test vs Bayesian two-sample test (00:28) 2. Confidence interval vs credible interval (02:10) 3. Bayes' theorem (04:15) 4. The prior distribution (09:20) 5. How to compute the Posterior distribution with simulations (12:47) 6. How to calculate the credible interval (17:33) 7. Prior * Likelihood (19:37) 8. The highest density interval (HDI) (23:00) 9. How to compute the p-...
How to check normal distribution | The normality assumption
Просмотров 9315 месяцев назад
See all my videos at: www.tilestats.com/ 1. Histogram 2. QQ plot (02:45) 3. Shapiro-Wilk (03:11) 4. An example of exponential distribution (08:40) 5. Type 1 and 2 errors in Shapiro-Wilk (10:49) 6. The normality assumption (14:49) 7. How to check the normality assumption (15:43)
PERMANOVA and permutation tests - explained
Просмотров 2,5 тыс.6 месяцев назад
www.tilestats.com/ 1. How to calculate the t-statistic 2. Permutation tests (02:40) 3. PERMANOVA (07:25)
Statistical power - Parametric vs Nonparametric test
Просмотров 1 тыс.7 месяцев назад
See all my videos at: www.tilestats.com/ In this video we will discuss the differences in statistical power and type 1 errors between the t-test and the Mann-Whitney U test (Wilcoxon Mann-Whitney (WMW), Wilcoxon rank-sum test) 1. Type 1 error - t-test vs WMW 2. Type 2 error - t-test vs WMW - normal distribution (04:30) 3. Type 2 error - t-test vs WMW - log-normal distribution (10:56) 4. R code ...
Meta-analysis | The inverse variance method | Forest plot in R
Просмотров 1,2 тыс.7 месяцев назад
www.tilestats.com/ 1. Meta-analysis - the basics 2. The inverse variance method (01:30) 3. How to interpret a forest plot (07:51) 4. Fixed vs random model (08:55) 5. How to create a forest plot in R (09:32)
The Mantel-Haenszel method - clearly explained | deal with confounding
Просмотров 3,2 тыс.8 месяцев назад
www.tilestats.com/ 1. The MH RR 2. The MH OR (06:52)
Understanding the odds ratio (OR) and the rare disease assumption | OR = RR?
Просмотров 7398 месяцев назад
www.tilestats.com/ 1. A case-control study - example: smoking and lung cancer 2. How to calculate the OR (01:00) 3. How to interpret the OR (03:46) 4. How to calculate the 95% confidence interval of the OR (04:47) 5. OR = RR ? (06:25) 6. The rare disease assumption (08:18)
Relative risk - how to calculate and interpret | 95% CI
Просмотров 1,7 тыс.8 месяцев назад
www.tilestats.com/ 1. Smokers vs non-smokers - lung cancer 2. How to calculate the RR (01:10) 3. How to interpret the RR (02:40) 4. How to calculate the 95% confidence interval of the RR (04:00) 5. How to interpret 95% confidence intervals (06:20) 6. Interpret the RR in relation to the actual risk (08:10)
Odds vs Probability - explained
Просмотров 7968 месяцев назад
Odds vs Probability - explained
The SIR model | the math of epidemics - explained with a simple example
Просмотров 5308 месяцев назад
The SIR model | the math of epidemics - explained with a simple example
Receptor ligand kinetics | mathematical modeling
Просмотров 5469 месяцев назад
Receptor ligand kinetics | mathematical modeling
How to build a system of differential equations (ODEs)
Просмотров 6919 месяцев назад
How to build a system of differential equations (ODEs)
How to select a multivariate analysis or machine learning method
Просмотров 3,3 тыс.9 месяцев назад
How to select a multivariate analysis or machine learning method
How to solve ordinary differential equations (ODEs) in R (deSolve)
Просмотров 1,6 тыс.10 месяцев назад
How to solve ordinary differential equations (ODEs) in R (deSolve)
Euler's method | Numerial methods
Просмотров 53110 месяцев назад
Euler's method | Numerial methods
Understanding ordinary differential equations (ODE) - super simple example
Просмотров 67211 месяцев назад
Understanding ordinary differential equations (ODE) - super simple example
The Cox proportional hazards model explained
Просмотров 15 тыс.11 месяцев назад
The Cox proportional hazards model explained
Comparing Kaplan-Meier curves - the Log-rank test
Просмотров 3,4 тыс.Год назад
Comparing Kaplan-Meier curves - the Log-rank test
Kaplan Meier curve - explained
Просмотров 8 тыс.Год назад
Kaplan Meier curve - explained
Nonlinear mixed effects models (NLME) - explained
Просмотров 6 тыс.Год назад
Nonlinear mixed effects models (NLME) - explained
Nonlinear regression - how to fit a logistic growth model to data
Просмотров 1,6 тыс.Год назад
Nonlinear regression - how to fit a logistic growth model to data
Nonlinear regression - how to fit a dose-response curve in R
Просмотров 3,3 тыс.Год назад
Nonlinear regression - how to fit a dose-response curve in R
Nonlinear regression - comparing models with F test and AIC | parameter correlation
Просмотров 1,9 тыс.Год назад
Nonlinear regression - comparing models with F test and AIC | parameter correlation

Комментарии

  • @dawitabathun9402
    @dawitabathun9402 День назад

    great

  • @dawitabathun9402
    @dawitabathun9402 День назад

    great

  • @cakefactoryy
    @cakefactoryy День назад

    Great video; it helped me alot!🔥

  • @asosalih257
    @asosalih257 День назад

    I was in need for this and for my surprise you have make one... thank you sir you always the great

  • @caiobustani5223
    @caiobustani5223 2 дня назад

    This is some very good material! Thanks for the effort of making such clear and perfectly paced explanation!

  • @RevisionNotes_hcleelars
    @RevisionNotes_hcleelars 5 дней назад

    Very good summary

  • @supa.scoopa
    @supa.scoopa 6 дней назад

    Wow. Thank you so much

  • @mgpetrus
    @mgpetrus 6 дней назад

    Thanks for your very didatical demostration. I was wondering why you didn't mentioned about the data transformation and the data standarlization previous start the analysis, mainly because the blood preasure and body size have distinct scales.

    • @tilestats
      @tilestats 5 дней назад

      Yes, you can standardize the data but you will get the same correlations with un-standardized data because you later on instead standardize the scores as I explain at 10:56.

  • @sakkariyaibrahim2650
    @sakkariyaibrahim2650 7 дней назад

    Great. After wandering through hundreds of video now I know what an Eigen vector and Eigen value is👍

  • @stonecastle858
    @stonecastle858 8 дней назад

    Excellent explanation, thank you

  • @ogbonnaya1004
    @ogbonnaya1004 8 дней назад

    Thank you, bro. Well explained. Love it😊

  • @sofiatrifonova5860
    @sofiatrifonova5860 10 дней назад

    Pairrrrssss 🥸 pehrz 😌

  • @user-yp1rg2jr5z
    @user-yp1rg2jr5z 12 дней назад

    How can we estimate the parameters of this model? Can we just use ols method by using the linear model (b+b1.x)? Which is used as power of "e" here?

    • @tilestats
      @tilestats 12 дней назад

      No, have a look at this video: ruclips.net/video/J0yuLu3oLuU/видео.html

  • @VivekGupta-sh3lj
    @VivekGupta-sh3lj 13 дней назад

    Huge thanks Can we say the pseudo-r-square is the same as deviance ratio that is reported in some statistical packages after logit models

  • @user-yp1rg2jr5z
    @user-yp1rg2jr5z 14 дней назад

    But why call it poisson regression where the graph you used is clearly follows a exponential distribution?

    • @tilestats
      @tilestats 14 дней назад

      Because the data points around the fitted curve follow a Poisson distribution.

  • @Winbugs1
    @Winbugs1 14 дней назад

    Very nice explanation

  • @AMADIGIFT-ui8ud
    @AMADIGIFT-ui8ud 16 дней назад

    This is so great. You're a good teacher, you explained what I have been trying to understand for more than a year in less than 30 minutes. You just got a new subscriber.

  • @user-yp1rg2jr5z
    @user-yp1rg2jr5z 17 дней назад

    Please make about linear approximation method before applying Least squares method.

  • @nguyentungle2183
    @nguyentungle2183 18 дней назад

    Your videos are great. You know that your video are one of learning materials for Statistics course in University of Groningen right? Much thanks

    • @tilestats
      @tilestats 18 дней назад

      No i did not know that but it is great if it is useful.

  • @ritiksuri7248
    @ritiksuri7248 18 дней назад

    Can you please provide the dataset on which you worked

    • @tilestats
      @tilestats 18 дней назад

      The data set is the same as shown in the video.

  • @richardgordon
    @richardgordon 19 дней назад

    Thank you for such a well explained video on Bootstrapping! Really impressive and clear!

  • @MrAwm-ki6jb
    @MrAwm-ki6jb 19 дней назад

    Really easy to understand. Cause you explain the reason why. Where i can't get this from my teacher.

  • @georgeyandem8629
    @georgeyandem8629 19 дней назад

    You are the best in my opinion. And I'm not bluffing

  • @veniasblack
    @veniasblack 20 дней назад

    Amazing explanation. Thanks alot

  • @flexogore4614
    @flexogore4614 20 дней назад

    What if the correlation coefficient equals to 1? In that case we would receive 0 in denominator under the square root sign. Overall, thanks a lot for this insightful course

    • @tilestats
      @tilestats 20 дней назад

      True, then the t-stat will be infinity large.

    • @flexogore4614
      @flexogore4614 20 дней назад

      @@tilestatsgiven we can't divide by 0, is it then replaced by some default infinitely small number?

  • @minhhai4318
    @minhhai4318 21 день назад

    this video is a gem .

  • @stephensonal4082
    @stephensonal4082 21 день назад

    Hi, you lost me for a moment at 5.42/13.11 where you calculated upper and lower CL - could you please layout further, how you achieved 43.37

  • @olalekeoluwaseyi9449
    @olalekeoluwaseyi9449 23 дня назад

    Wow, this is awesome and very insightful I must subscribe to your channel

  • @learning_with_irving4266
    @learning_with_irving4266 23 дня назад

    How are the 2.747 and 5.7 derived?

    • @tilestats
      @tilestats 22 дня назад

      That is explained at 11:30 and forward.

  • @khaingzar3136
    @khaingzar3136 23 дня назад

    Which software is used to get the equation for model Price = constant + Age.Coefficient + Mileage.Coefficient ?

    • @tilestats
      @tilestats 22 дня назад

      You have to create the equation on your own and then use the software to estimate the parameter values for the equation. I use R but you can use any other statistical software to get the same parameter values.

  • @mayling1014
    @mayling1014 26 дней назад

    Thank you for the detailed explanation! 1:12 May I know if both adjusting the significance level and adjusting p value result in the same conclusion, how do we decide which one to use? I see most of the papers use adjusted p value...

    • @tilestats
      @tilestats 25 дней назад

      You will come to the same conclusion. In a paper, it might be confusing if you use different significance levels in different test. It is therefore easier if you use just one alpha (usually 0.05) and adjust the p-values.

    • @mayling1014
      @mayling1014 25 дней назад

      @@tilestats Understood :) Thank you so much for your prompt reply. May I ask 2 questions? 1) Is the term "FDR adjusted p-value" interchangeable with "q-value"? 2) For RNA sequencing, I have significant DEGs when using p<0.05, but no significant DEGs when using adjusted p <0.05. Can I still use p value in order to get DEGs for downstream biological annotation?

    • @tilestats
      @tilestats 24 дня назад

      FDR adjusted p-values usually refer to BH adjusted p-values or q-values. To see the difference, watch this video: ruclips.net/video/T6J4b-WWebM/видео.html You can still use, for example, GSEA with a ranked gene list based on log2FC as I show in this video: ruclips.net/video/EF94wPaqXM0/видео.html

  • @Mathclub63
    @Mathclub63 26 дней назад

    Absolutely ❤❤❤❤

  • @fallenangel8785
    @fallenangel8785 26 дней назад

    One of the Most underrated channels in RUclips

  • @surendrabarsode8959
    @surendrabarsode8959 26 дней назад

    Very well explained. Anyone can easily very well understand these concepts. Thanks!!!

  • @miguelchiri8895
    @miguelchiri8895 26 дней назад

    What book can I find this in?

    • @tilestats
      @tilestats 26 дней назад

      I have not used any book but you can buy most of the videos as PDFs on my home page: www.tilestats.com

    • @miguelchiri8895
      @miguelchiri8895 26 дней назад

      @@tilestats I'm looking for bibliographic references on this topic but so far I haven't been able to find any. :(

  • @ast3362
    @ast3362 26 дней назад

    6:57 I don't understand how we can now assume that the Groups have one random intercept. That would mean each group has their own distribution, which is not the case because as you said all 4 subjects are randomly sampled out of one distribution.

    • @tilestats
      @tilestats 26 дней назад

      When we include Diet in the model, we can test if the two groups have different intercepts. Thus, we then no longer assume that all individuals are sampled from the same distribution.

    • @ast3362
      @ast3362 25 дней назад

      @@tilestats thanks! I just realised the difference between groups and clusters. Now everything is clear!

  • @CalzOmon
    @CalzOmon 26 дней назад

    Amazing! Thanks so much

  • @dawitmusse3548
    @dawitmusse3548 28 дней назад

    At 11:00 how are the weights optimised, please?

    • @tilestats
      @tilestats 27 дней назад

      Have a look at this video, where I explain how the weights are optimized: ruclips.net/video/XxZ0BibMTjw/видео.html

    • @dawitmusse3548
      @dawitmusse3548 27 дней назад

      @@tilestats Thank you so much, sir!

  • @lorenzotagliari6699
    @lorenzotagliari6699 28 дней назад

    I did not understand why the cutoff od 0.001 would not be appropriate in cases when we have many datapoints. Could you clear this up for me?

    • @tilestats
      @tilestats 27 дней назад

      Because, 0.1% of the data points will be outside the ellipse due to chance. If you for example have 1 million data points, you should expect that 1000 are outside the ellipse, right? It would then not be appropriate to define all these as outliers.

  • @sridharr2025
    @sridharr2025 28 дней назад

    Very useful video, thank you

  • @psychologykaTopper
    @psychologykaTopper Месяц назад

    In my research, there are two groups: control and experimental. Both of these groups gave pre- and post-tests. Which test should I use? In the experimental group, there are 100 people,50 male, 50 female and in the control group, there are 100 people.50 male, 50 female, Could you please explain how to calculate the mean and standard deviation for this large sample? Thank you!”

    • @tilestats
      @tilestats 29 дней назад

      You can use an unpaired t-test between exp group and control group based on the differences between pre and post test.

    • @psychologykaTopper
      @psychologykaTopper 29 дней назад

      @@tilestats thanku so much, If my population is larger, do I still need to use a paired t-test over a z-test?

    • @tilestats
      @tilestats 28 дней назад

      Z-test is generally only used if you know the population variance (which is usually not the case). For a large sample size, the Z-test and t-test will result in about the same p-value.

    • @psychologykaTopper
      @psychologykaTopper 28 дней назад

      @@tilestats thanku so much.

  • @user-yp1rg2jr5z
    @user-yp1rg2jr5z Месяц назад

    Here , 'Binom(6,8,p)' is this a likelihood or conditional distribution of x given parameter? [ f(x | p) ] I think both are same 🙂 or I don't know.

    • @tilestats
      @tilestats 29 дней назад

      It is the likelihood. When you are trying to estimate the parameter p based on observed data (6 successes out of 8 trials), you are using it as a likelihood function. You may watch this video to get a better understanding: ruclips.net/video/PRpmA6WsY6g/видео.html

  • @stevencaramoy2043
    @stevencaramoy2043 Месяц назад

    Such good explanation for the beginner like me

  • @morielgenish6515
    @morielgenish6515 Месяц назад

    thank you! if i want to calculate a confidence interval at other confidence levels, for example 98%?

    • @tilestats
      @tilestats Месяц назад

      Then you just use 98% instead of 95% in this video.

  • @FranciscoLozano-ov3zs
    @FranciscoLozano-ov3zs Месяц назад

    nice video! how do you do the log-rank test when you have several groups? in my particular case i have 13 groups. what can i do to reduce the family wise error rate?

    • @tilestats
      @tilestats Месяц назад

      If you want to do many pairwise comparisons, you can adjust your p-values by, for example, Bonferroni: ruclips.net/video/4_V2m41vpZw/видео.html or Holm's test: studio.ruclips.net/user/videol4yVt_Dht4U/edit

  • @harkatiyoussef9994
    @harkatiyoussef9994 Месяц назад

    by far one of the best, to not say the best explained on this subject (without exaggerating)

  • @gd8109
    @gd8109 Месяц назад

    Your videos give the best explanations! You make it so clear and easy to follow. Thank you!

  • @a.mo7a
    @a.mo7a Месяц назад

    why using 2 output nodes? isn't P(healthy) equal to 1-P(cancer)?

    • @tilestats
      @tilestats Месяц назад

      True, you can use just one output node when you predict just two categories. The R code I provided generates two output nodes but if you try TensorFlow in Python, it will use just one output if you set loss='binary_crossentropy'.

    • @a.mo7a
      @a.mo7a Месяц назад

      @@tilestats Thanks for the clarification

  • @rema2769
    @rema2769 Месяц назад

    Thank you so much, this is really helpful I finally understand now 24hr before my final

  • @anmolpardeshi3138
    @anmolpardeshi3138 Месяц назад

    I see that you centered the data. Is only centering required for "standardization" or scaling is also normally done such that the mean =0; standard deviation=1? this will then change the covariance matrix since variance of individual dimensions will equal 1.

    • @tilestats
      @tilestats Месяц назад

      It is not a requirement, mathematically, to standardize your data (mu = 0, SD = 1), but it is highly recommended, especially if you have variables with a large difference in the variance. I discuss that in the next video about PCA: ruclips.net/video/dh8aTKXPKlU/видео.html