Bootstrap is very simple technique used for small samples. Major takeaways from video are: What?: Resampling with replacement from sample data. Why?: To find std errors without invoking CLT. How?: K times repetitive sampling of size n (observation). When to use?: Most suitable for small sample sizes. Major Advantage: No need to invoke CLT or normality assumption
Views: 19134 Sarveshwar Inani
This screencast continues the discussion and tutorial of using the non-parametric bootstrap for statistical inference, in this case for regression models (and the general linear model more generally).
Views: 4761 Ian Dworkin
Using the R programming language to perform non-parametric bootstrap for statistical inferences, in particular generating confidence intervals. This includes the random variable ("pairs") bootstrap and the residual (fixed effect) bootstrap.
Views: 3600 Ian Dworkin
Bootstrapping and Resampling in Statistics with Example: What is Bootstrapping in Statistics and Why Do We Use it? 👉🏼Related Videos: Bootstrapping in Statistics & Bootstrapping in R Series: https://bit.ly/2GL6AYS 👍🏼Best Statistics & R Programming Language Tutorials: ( https://goo.gl/4vDQzT ) ►► Like to support us? You can Donate (https://bit.ly/2CWxnP2), Share our Videos, Leave us a Comment and Give us a Like! Either way We Thank You! ▶︎ In this statistics video lecture we will learn the Bootstrap method (a brute force method), along with why one may want to use such an approach. ▶︎ Bootstrap in statistics is a re-sampling based approach, useful for estimating the sampling distribution and standard error of an estimate. Bootstrapping provides an alternative approach to approaches based on large sample theory (you may recall that many approaches rely on having a large n in order to carry out the method). ▶︎ Bootstrapping in statistics and in research becomes particularly useful when dealing with more complicated estimates, where their standard error may not be easily calculated. ▶︎ This video takes a simple example, and explains the principle of a bootstrap approach, and the concept it is based on. The intention is to clarify exactly what this method involves. This video uses an estimate of a "sample mean" for simplicity of explaining the concept, although the bootstrap's real value comes when dealing with more complicated estimates...here, we simply aim to provide a conceptual understanding of this approach. ► ► Watch More: ► Intro to Statistics Course: https://bit.ly/2SQOxDH ►Data Science with R https://bit.ly/1A1Pixc ►Getting Started with R (Series 1): https://bit.ly/2PkTneg ►Graphs and Descriptive Statistics in R (Series 2): https://bit.ly/2PkTneg ►Probability distributions in R (Series 3): https://bit.ly/2AT3wpI ►Bivariate analysis in R (Series 4): https://bit.ly/2SXvcRi ►Linear Regression in R (Series 5): https://bit.ly/1iytAtm ►ANOVA Concept and with R https://bit.ly/2zBwjgL ►Hypothesis Testing: https://bit.ly/2Ff3J9e ►Linear Regression Concept and with R Lectures https://bit.ly/2z8fXg1 Follow MarinStatsLectures Subscribe: https://goo.gl/4vDQzT website: https://statslectures.com Facebook:https://goo.gl/qYQavS Twitter:https://goo.gl/393AQG Instagram: https://goo.gl/fdPiDn Our Team: Content Creator: Mike Marin (B.Sc., MSc.) Senior Instructor at UBC. Producer and Creative Manager: Ladan Hamadani (B.Sc., BA., MPH) These videos are created by #marinstatslectures to support some statistics and R programming language courses at The University of British Columbia (UBC) (#IntroductoryStatistics and #RVideoTutorials for Health Science Research), although we make all videos available to the everyone everywhere for free. #statistics #rprogramming
Views: 7230 MarinStatsLectures- R Programming & Statistics
Bootstrap Confidence Intervals in R with Example: How to build bootstrap confidence intervals in R without package? 👉🏼Link to Practice R Dataset (chickdata) & R-Script :(https://bit.ly/2rOfgEJ) 👍🏼Best Statistics & R Programming Language Tutorials: ( https://goo.gl/4vDQzT ) ►► Want to support us? You can Donate (https://bit.ly/2CWxnP2), Share Our Videos, Leave Comments or Give a us Like! What to Expect: ►In this R video tutorial, we learn how to use R programming language to generate a confidence interval using a bootstrap approach, Step by Step with no R Package ► In this R tutorial we will focus on comparing the means (and medians) of two different groups or samples ► R packages do exist for bootstrapping in R (one package name: boot), although the package is limited in the sorts of estimates/statistics it can conduct a bootstrap approach for. Our goal is to show you how to build the bootstrap approach yourself, so that you can change the sorts of statistics/estimates you can work with. ► You can always practice building the interval yourself, and then comparing the results to what you get using the "boot" package in R. Note that if you do this, numeric values will differ slightly because you and the package will end up with a different set of bootstrap samples, and so there will be a slight numeric difference in results. General Overview: ► Bootstrapping in statistics is a resampling based approach useful for estimating the sampling distribution and standard error of an estimate ► Bootstrapping provides an alternative approach to approaches based on large-sample theory (you may recall that many approaches rely on having a large n in order to carry out the method). ► Bootstrapping becomes particularly useful when dealing with more complicated estimates, where their standard error may not be easily calculated, or the shape of their sampling distribution is not easy to estimate ► Many classic approaches make inferences about means/medians, etc. Bootstrapping opens the door to working with much more interesting, and informative, estimates ■Table of Content: Coming Soon ►►Watch More: ► Bootstrapping in Statistics and Bootstrapping in R Videos: https://bit.ly/2GL6AYS ►Bootstrapping and Resampling Video: https://youtu.be/O_Fj4q8lgmc ►Bootstrap Hypothesis Testing Video: https://youtu.be/9STZ7MxkNVg ►Bootstrap Hypothesis Testing in R: https://youtu.be/Zet-qmEEfCU ►Two Sample t Test in R: https://goo.gl/rVzjyf ►Two Sample t Test Concept: https://youtu.be/mBiVCrW2vSU ►Mann Whitney U aka Wilcoxon Rank-Sum Test in R https://youtu.be/KroKhtCD9eE Follow MarinStatsLectures Subscribe: https://goo.gl/4vDQzT website: https://statslectures.com Facebook: https://goo.gl/qYQavS Twitter: https://goo.gl/393AQG Instagram: https://goo.gl/fdPiDn Our Team: Content Creator: Mike Marin (B.Sc., MSc.) Senior Instructor at UBC. Producer and Creative Manager: Ladan Hamadani (B.Sc., BA., MPH) These videos are created by #marinstatslectures to support some statistics courses at the University of British Columbia (UBC) (#IntroductoryStatistics and #RVideoTutorials ), although we make all videos available to the everyone everywhere for free. Thanks for watching! Have fun and remember that statistics is almost as beautiful as a unicorn! 🦄 #statistics #rprogramming
Views: 1617 MarinStatsLectures- R Programming & Statistics
In this video, I demonstrate how to write a for loop in R to perform a simple bootstrap calculation. My working example is computing a bootstrap approximation to the sampling distribution for R-squared from a simple linear regression model. I embed all of my R video tutorials on my econometrics blog here: http://novicemetrics.blogspot.com If you want to download RStudio because you like the way it is organized relative to R, you can do so at www.RStudio.org. A description of this video and the code I used with comments is available here: http://novicemetrics.blogspot.com/2011/03/how-to-bootstrap-in-r-case-of-r-squared.html
Views: 20783 intromediateecon
In this video, I explain how to make a confidence interval with bootstrapping. Here, we go over how to make a confidence interval with the true population, how to apply bootstrap to get the confidence interval and finally, I walk you through what happens to the confidence interval as the sample size increases and decreases. Enjoy! Link to my notes on Introduction to Data Science: https://github.com/knathanieltucker/data-science-foundations Try answering these comprehension questions to further grill in the concepts covered in this video: 1. What is the central limit theorem? And what is the classic way of making confidence intervals? 2. When can we apply bootstrap confidence intervals? 3. What is the sampling distribution and why is it special? 4. Is it reasonable to assume that our sampling distribution will be symmetric? 5. What happens to the sampling distribution as the original sample size increases? As the number of samples to create it increases? And what happens to the bootstrap sampling distribution? 6. What is annoying about the first bootstrap confidence interval that was taught? 7. Can you show empirically that the percentile confidence interval works? Code it. 8. Is there a way that you can estimate the percent of times your confidence interval will be right without the population distribution?
Views: 388 Data Talks
This video is going to show how to perform cross validation and bootstrapping in R. Thanks for watching. My website: http://allenkei.weebly.com If you like this video please "Like", "Subscribe", and "Share" it with your friends to show your support! If there is something you'd like to see or you have question about it, feel free to let me know in the comment section. I will respond and make a new video shortly for you. Your comments are greatly appreciated.
Views: 2455 Allen Kei
Lecture with Per B. Brockhoff. Lecture 10. Chapters: 00:00 - Two Samples; 02:15 - Example 3; 03:45 - Example 3, Solution In R;
Views: 1869 DTUdk
In this talk, presented at UseR! 2016, I will explain the model behind the Bayesian bootstrap, how it connects to the classical bootstrap and in what situations the Bayesian bootstrap is useful. I will also show how one can easily perform Bayesian bootstrap analyses in R using my package bayesboot (https://cran.r-project.org/package=bayesboot). For more info on the Bayesian bootstrap see my blog: http://www.sumsar.net/blog/2015/04/the-non-parametric-bootstrap-as-a-bayesian-model/
Views: 2673 rasmusab
Lecturer: Dr. Erin M. Buchanan Missouri State University Summer 2017 This lecture + demonstration shows you how to do mediation analyses, the Sobel test, and bootstrapping of the indirect effect in R. I compare the code in this demonstration to the output provided from the Process plug-in in SPSS. Lecture materials and assignment available at statstools.com. http://statstools.com/learn/advanced-statistics/
Views: 2966 Statistics of DOOM
Learn more about credit risk modeling in R: https://www.datacamp.com/courses/introduction-to-credit-risk-modeling-in-r We have seen several techniques for preprocessing the data. When the data is fully preprocessed, you can go ahead and start your analysis. You can run the model on the entire data set, and use the same data set for evaluating the result, but this will most likely lead to a result that is too optimistic. One alternative is to split the data into two pieces. The first part of the data, the so-called training set, can be used for building the model and the second part of the data, the test set, can be used to test the results. One common way of doing this is to use two-thirds of the data for a training set and one-third of the data for the test set. Of course there can be a lot of variation in the performance estimate depending which two-thirds of the data you select for the training set. One way to reduce this variation is by using cross validation. For the two-thirds training set and one-third test set example, a cross validation variant would look like this. The data would be split in three equal parts, and each time, two of these parts would act as a training set, and one part would act as a test set. Of course, we could use as many parts as we want, but we would have to run the model many times if using many parts. This may become computationally heavy. In this course, we will just use one training set and one test set containing two-thirds versus one-third of the data, respectively. Imagine we have just run a model, and now we apply the model to our test set to see how good the results are. Evaluating the model for credit risk means comparing the observed outcomes of default versus non-default--stored in the loan_status variable of the test set--with the predicted outcomes according to the model. If we are dealing with a large number of predictions, a popular method for summarizing the results uses something called a confusion matrix. Here, we use just 14 values to demonstrate the concept. A confusion matrix is a contingency table of correct and incorrect classifications. Correct classifications are on the diagonal of the confusion matrix. We see, for example, that 8 non-defaulters were correctly classified as non-default, and 3 defaulters were correctly classified as defaulters. However, we see that 2 non-defaulters where wrongly classified as defaulters, and 1 defaulter was wrongly classified as a non-defaulter. The items on the diagonals are also called the true positives and true negatives. The off-diagonals are called the false positives versus the false negatives. Several measures can be derived from the confusion matrix. We will discuss the classification accuracy, the sensitivity and the specificity. The classification accuracy is the percentage of correctly classified instances, which is equal to 78.57% in this example. The sensitivity is the percentage of good customers that are classified correctly, or 75% in this example. The specificity is the percentage of bad costomers that are classified correctly, or 0.80 in this example. Let's practice splitting the data and constructing confusion matrices.
Views: 15632 DataCamp
Machine Learning #49 Bootstrapping & Cross Validation Machine Learning Complete Tutorial/Lectures/Course from IIT (nptel) @ https://goo.gl/AurRXm Discrete Mathematics for Computer Science @ https://goo.gl/YJnA4B (IIT Lectures for GATE) Best Programming Courses @ https://goo.gl/MVVDXR Operating Systems Lecture/Tutorials from IIT @ https://goo.gl/GMr3if MATLAB Tutorials @ https://goo.gl/EiPgCF
Views: 6106 Xoviabcs
Bootstrapping uses the observed data to simulate resampling from the population. This produces a large number of bootstrap resamples. We can calculate a statistic for each bootstrap resample and use the distribution of the simulated statistics to approximate characteristics of the population. This bootstrapping process can help us construct a confidence interval for a population parameter, even when the population distribution is unknown. This video illustrates how to calculate a bootstrap confidence interval for regression coefficients.
Views: 2911 Professor Knudson
Lecture with Per B. Brockhoff. Lecture 10. Chapters: 00:00 - Introduction; 01:45 - Confidence Intervals Using Simulation: Bootstrapping; 06:00 - Non-Parametric Bootstrap For The One-Sample Situation; 09:00 - Example 2, One-Sample; 11:15 - Example 2, Solution In R;
Views: 7249 DTUdk
In this video we are going to be applying our knowledge on bootstrap to regression in a regression bootstrap example where we explore the relationship between sports and hydration.
Views: 279 Data Talks
A tutorial on how to use R to create bootstrap confidence intervals for a single proportion (using both the plug-in and percentile methods).
Views: 4276 LawrenceStats
Thisis a way to test the model. It is a random sample conducted with replacement. Lets consider to have many variables: age, gender, income. So, to generate bootstrap sample we have to randomly select a sample from the original data, and we continue this process for a number of times. The final result is one bootstap sample with N observation that we can use as test set or on the contrary we can generate a series of training sets. Bootsrapping can be also used to egenrate a confidence interval for metrics when confidence interval s are difficult to estimate (a confidence interval is a range of values we fairly sure our true value lies in). Lets' say you want thevariable median but you have few observations: 1 - take a boostrap sample 2 - compute the median 3 - repeat a large number of times This procedures produces an empirical distribution of the median from which you can calcualte the confidence interval and test statistical hypothesis without to assume a spesific underlying theoretical distribution. There is a great R package to do that which is called: boot package, and the boot.ci() function is used to obtain confidence intervals for the statistics needed (e.g media, R-squared, coeficients of a model). In R: there is a great R package to do that which is called: boot package, and the boot.ci () function is used to obtain confidence intervals for the statistics that we need to estimate (e.g. median, R-squared, coefficients of a model).
Views: 39 Science as falsification
decision Tree, though a good model to interpret the outcomes, not a good model in terms of prediction accuracy. Bagging is a combination of several decision trees that helps in improving the prediction accuracy of decision tree model. It can be used for both regression & classification problems Contact :[email protected] ANalytics Study Pack : https://analyticuniversity.com/ Analytics University on Twitter : https://twitter.com/AnalyticsUniver Analytics University on Facebook : https://www.facebook.com/AnalyticsUniversity Logistic Regression in R: https://goo.gl/S7DkRy Logistic Regression in SAS: https://goo.gl/S7DkRy Logistic Regression Theory: https://goo.gl/PbGv1h Time Series Theory : https://goo.gl/54vaDk Time ARIMA Model in R : https://goo.gl/UcPNWx Survival Model : https://goo.gl/nz5kgu Data Science Career : https://goo.gl/Ca9z6r Machine Learning : https://goo.gl/giqqmx Data Science Case Study : https://goo.gl/KzY5Iu Big Data & Hadoop & Spark: https://goo.gl/ZTmHOA
Views: 7138 Analytics University
All of my videos use "annotations." Make sure that you have annotations turned on or you might miss important information, such as error correction! You can make sure annotations are on by clicking on the gear-shaped symbol near the bottom-right corner of the player. I will never use annotations to for advertising or self-promotion. Link to R Statistical Software homepage: http://www.r-project.org/ Link to RStudio homepage: http://www.rstudio.com/ Link to Lavaan homepage: http://lavaan.ugent.be/ Link to Dr. Yves Rosseel's Lavaan Tutorials: http://lavaan.ugent.be/tutorial/index.html DESCRIPTION This video will walk you through path analysis using the "Lavaan" package in R. I cover the basic steps to estimate model parameters as well as the additional steps needed to estimate indirect effects. NOTE ABOUT SIGNIFICANCE TESTS By default, Lavaan provides significance tests for most effects based on the assumption that the sampling distributions of those effects are normally distributed. There are many cases in which this assumption is not supported (e.g., indirect effects) and you might wish to use an alternative method for significance testing, such as bootstrapped confidence intervals. Lavaan is capable of providing results from these alternative procedures, but a discussion of this topic goes beyond the scope of this video. Although I plan to make separate videos to discuss these methods, I am available to help if you have a specific need. All I ask is that you use the comment section below so that others might benefit from your question. For more info about significance tests for indirect effects in path analysis: Hayes, A. F., & Scharkow, M. (2013). The relative trustworthiness of inferential tests of the indirect effect in statistical mediation analysis does method really matter?. Psychological Science, 0956797613480187. ABOUT MY LAVAAN TUTORIALS These tutorials are based heavily on the work conducted by the developers of Lavaan. The main developer, Yves Rosseel at Ghent University, has put together an excellent set of tutorials (available in PDF format) that go into the topics discussed on this channel in much more detail. I highly recommend this text for educational purposes and as a reference while you conduct your analyses (see link above). -------------------------------------- Let me know how I can improve future videos by leaving a comment!
Views: 25638 Jordan Clark
Bagging is the not-so-secret edge of the competitive modeler. By sampling and modeling a training data set hundreds of times and averaging its predictions, you may just get that accuracy boost that puts you above the fray. Walkthrough/code: http://amunategui.github.io/bagging-in-R/ MORE: Signup for my newsletter and more: http://www.viralml.com Connect on Twitter: https://twitter.com/amunategui My books on Amazon: The Little Book of Fundamental Indicators: Hands-On Market Analysis with Python: Find Your Market Bearings with Python, Jupyter Notebooks, and Freely Available Data: https://amzn.to/2DERG3d Monetizing Machine Learning: Quickly Turn Python ML Ideas into Web Applications on the Serverless Cloud: https://amzn.to/2PV3GCV Grow Your Web Brand, Visibility & Traffic Organically: 5 Years of amunategui.github.Io and the Lessons I Learned from Growing My Online Community from the Ground Up: Fringe Tactics - Finding Motivation in Unusual Places: Alternative Ways of Coaxing Motivation Using Raw Inspiration, Fear, and In-Your-Face Logic https://amzn.to/2DYWQas Create Income Streams with Online Classes: Design Classes That Generate Long-Term Revenue: https://amzn.to/2VToEHK Defense Against The Dark Digital Attacks: How to Protect Your Identity and Workflow in 2019: https://amzn.to/2Jw1AYS CATEGORY:DataScience HASCODE:True
Views: 9289 Manuel Amunategui
This is lecture 13 of the coursera class Statistical Inference. The lecture notes can be found here https://github.com/bcaffo/courses/blob/master/06_StatisticalInference/13_Resampling/index.pdf?raw=true Watch the full playlist here http://www.youtube.com/playlist?list=PLpl-gQkQivXiBmGyzLrUjzsblmQsLtkzJ
Views: 17280 Brian Caffo
Provides steps for carrying handling class imbalance problem when developing classification and prediction models Download R file: https://goo.gl/ns7zNm data: https://goo.gl/d5JFtq Includes, - What is Class Imbalance Problem? - Data partitioning - Data for developing prediction model - Developing prediction model - Predictive model evaluation - Confusion matrix, - Accuracy, sensitivity, and specificity - Oversampling, undersampling, synthetic sampling using random over sampling examples predictive models are important machine learning and statistical tools related to analyzing big data or working in data science field. R is a free software environment for statistical computing and graphics, and is widely used by both academia and industry. R software works on both Windows and Mac-OS. It was ranked no. 1 in a KDnuggets poll on top languages for analytics, data mining, and data science. RStudio is a user friendly environment for R that has become popular.
Views: 16187 Bharatendra Rai
In this video you will learn the theory behind a random forest model. Random forest is a type of ensemble model that uses multiple weak classifier (decision tree) to build a strong classifier. It is very similar to a bagging model but the difference is that it adds additional layer of randomness by choosing predictors at random unlike bagging that uses the same set of predictors (all of the available) for splitting the data. It is a powerful machine learning model. Contact us at @ [email protected] ANalytics Study Pack : http://analyticuniversity.com/ Bagging model : https://www.youtube.com/watch?v=ab3IcEPwa3A Analytics University on Twitter : https://twitter.com/AnalyticsUniver Analytics University on Facebook : https://www.facebook.com/AnalyticsUniversity Logistic Regression in R: https://goo.gl/S7DkRy Logistic Regression in SAS: https://goo.gl/S7DkRy Logistic Regression Theory: https://goo.gl/PbGv1h Time Series Theory : https://goo.gl/54vaDk Time ARIMA Model in R : https://goo.gl/UcPNWx Survival Model : https://goo.gl/nz5kgu Data Science Career : https://goo.gl/Ca9z6r Machine Learning : https://goo.gl/giqqmx Data Science Case Study : https://goo.gl/KzY5Iu Big Data & Hadoop & Spark: https://goo.gl/ZTmHOA
Views: 15810 Analytics University
Learn more about machine learning with R: https://www.datacamp.com/courses/machine-learning-toolbox In the last video, we manually split our data into a single test set, and evaluated out-of-sample error once. However, this process is a little fragile: the presence or absence of a single outlier can vastly change our out-of-sample RMSE. A better approach than a simple train/test split is using multiple test sets and averaging out-of-sample error, which gives us a more precise estimate of true out-of-sample error. One of the most common approaches for multiple test sets is known as "cross-validation", in which we split our data into ten "folds" or train/test splits. We create these folds in such a way that each point in our dataset occurs in exactly one test set. This gives us 10 test sets, and better yet, means that every single point in our dataset occurs exactly once. In other words, we get a test set that is the same size as our training set, but is composed of out-of-sample predictions! We assign each row to its single test set randomly, to avoid any kind of systemic biases in our data. This is one of the best ways to estimate out-of-sample error for predictive models. One important note: after doing cross-validation, you throw all resampled models away and start over! Cross-validation is only used to estimate the out-of-sample error for your model. Once you know this, you re-fit your model on the full training dataset, so as to fully exploit the information in that dataset. This, by definition, makes cross-validation very expensive: it inherently takes 11 times as long as fitting a single model (10 cross-validation models plus the final model). The train function in caret does a different kind of re-sampling known as bootsrap validation, but is also capable of doing cross-validation, and the two methods in practice yield similar results. Lets fit a cross-validated model to the mtcars dataset. First, we set the random seed, since cross-validation randomly assigns rows to each fold and we want to be able to reproduce our model exactly. The train function has a formula interface, which is identical to the formula interface for the lm function in base R. However, it supports fitting hundreds of different models, which are easily specified with the "method" argument. In this case, we fit a linear regression model, but we could just as easily specify method = 'rf' and fit a random forest model, without changing any of our code. This is the second most useful feature of the caret package, behind cross-validation of models: it provides a common interface to hundreds of different predictive models. The trControl argument controls the parameters caret uses for cross-validation. In this course, we will mostly use 10-fold cross-validation, but this flexible function supports many other cross-validation schemes. Additionally, we provide the verboseIter = TRUE argument, which gives us a progress log as the model is being fit and lets us know if we have time to get coffee while the models run. Let's practice cross-validating some models.
Views: 48256 DataCamp
This video demonstrates how to perform bootstrapping in Lavaan to obtain bootstrap standard errors for indirect effects to test mediation (in addition to standard errors for direct and total effects). A copy of the text file containing the syntax used can be downloaded here: https://drive.google.com/open?id=1Qaus_aM8wmiT7-ZVHLsCNP8SzXStegEG A copy of the data can be downloaded here: https://drive.google.com/open?id=19phWbbt4YTckwhqG-LDpz6LJD5SDHCJK
Views: 52 Mike Crowson
Bootstrapping uses the observed data to simulate resampling from the population. This produces a large number of bootstrap resamples. We can calculate a statistic for each bootstrap resample and use the distribution of the simulated statistics to approximate characteristics of the population. This video lays the foundation for later bootstrap videos.
Views: 18096 Professor Knudson
Greta Cutulenco's talk at Waterloo Data Science meetup, November 19, 2015, Engineering 5 building, University of Waterloo, Ontario, Canada http://www.meetup.com/Waterloo-Data-Science/events/226637424/ Meetup's first part on Ensemble Learning is here https://youtu.be/cBtIGkH9SvA Watch also Greta's next month talk on Linear Modeling. here https://youtu.be/QYIX99lO5Tk Video starts & ends with outside night shots of University of Waterloo Engineering 5 building and surroundings. Recorded & edited by @curelet (http://ghcb.blogspot.ca www.linkedin.com/in/gcurelet) with Canon Powershot SX 520 HS superzoom point-and-shoot camera.
Views: 21654 Gheorghe Curelet-Balan