Details

Published: Saturday, 26 September 2020 22:02

Hits: 1243

Hair Cutting:

One thing that I've always considered very expensive in USA is a hair cut. They cost about $60/hr (assuming $15 haircut in 15 minutes). Of course, it includes shop cost and the hair cutter themselves make barely anything. Even the hair franchisee doesn't make too much (the owner who operates the hair salon). It's mostly the franchisor who makes most of the profit. Most of the hair salon chain are franchisor, i.e greatclips, supercuts are examples of franchisor who trade on stock markets.

Hair cut Clipper settings:

Clipper run from 1 to 27 or something. Size 1 cuts the smallest hair, while higher numbered clipper leaves your hair longer. 1 implies size of 1mm i.e that clipper only leaves 1 mm length of hair on your head, while size of 6 implies size of 6mm.

From the consumer side also, going for a hair cut is a huge wastage of time and money. He/she has to drive somewhere, spend a lot of time sitting there, then someone cuts hair for 15 min, then drive back and take shower, which easily results in more than an hour of time for a 10 min job.

So, essentially hair cutting is a very inefficient process from both consumer and producer side. Cutting your own hair is a much more efficient process. A good quality hair cutter will cost you about $10 and will last for couple of years. However, you do need another person to cut your hair, as cutting your own hair isn't easy (atleast not for me).

1. Hair Salon: If you still don't want to invest in cutting your own hair, these salons may provide a relatively cheaper option to get your hair cut.

• Great Clips: www.greatclips.com offers the cheapest option. You usually get coupons for $6 to $8 for basic haircut. If you don't get their coupons, you would want to sign up on their website.
• Local Hair Salon: Other great place is www.groupon.com. You can search under local deals, and can usually get a hair cut for under $10 (by applying 20% off coupon for local deals that groupon has almost every month).
• Training Institutes: These are hair styling training institutes that you'll have to search online. There are usually a few in any city. They cut hair for $5 or so. Hair cut quality is good, as they always have an instructor who is guiding them.

2. Hair Cutter/Trimmers: These are few hair cutters or trimmers that i've tried and they have worked well. there is no reason to spend anything more than $20 for any hair trimmer, as the expensive ones just have more bells and whistles, which aren't needed.

• Braun hair trimmer: https://www.amazon.com/gp/product/B077HQ6WH4. I bought this for $20.It's cordless, and battery lasts for multiple hair cuts.
  
  • Use the Long Beard trimmer (#3) with setting of 17 to go over all of your head.
  • Then for close cut, use short beard trimmer (#2) with setting of 7 to go around your neck and ears.
  • There is precision trimmer (#4) that you use around your ears and sideburns, as it gets close to skin to remove all small hair.
  • There is stubble look trimmer (#1) with 2 other settings of 1 and 2 that you should never use. It cuts your hair so short, that your skin is visible. So, if you accidently end up cutting hair at a wrong spot with these, there is no way to recover. It also looks very weird on that spot as hair is basically gone. The lowest you should go is may be setting of 3 on beard trimmer, and that also on very bottom hairs around neck and ears. Safest way is to stick with setting of 7, then very carefully use setting of 5 if still needed around the edges. Setting of 3 should be avoided too, until you get experienced using these clippers.

While cutting your hair is hard, cutting others hair is not that difficult. So, you can provide reciprocal services to your spouse, room mates, etc. Watch a few basic hair cutting videos on youtube, and practise it while watching. One key skill to cutting with clippers is that clippers need to go from the edges of your head to the center of your hear, i.e clipper needs to pierce thru your hair. If you take your clipper from the center of your head, to edges of your head, then you will notice that clipper is just sliding on your hair without cutting hair. If hair is not getting cut, you are using the clipper in wrong way. Watch those videos again.

Practice your skills on your kids without fears of adverse repercussions. This is one rare place where having lots of kids helps you. After all, their hair do grow fast !!

Details

Published: Sunday, 20 September 2020 16:37

Hits: 813

GDP : One way to calculate GDP is to count all the expenses incurred by people, businesses and Government. Here we look at the component of GDP pertaining to expenses incurred by consumers.

The bea.gov website has all the details of GDP. If we look at PCE (Personal Consumption Expenditure) for 2019, we see $14.8T  as total money spent by Consumers.

https://apps.bea.gov/iTable/iTable.cfm?ReqID=19&step=2#reqid=19&step=2&isuri=1&1921=underlying

We will divide consumer expenses in 2 parts => Retail sales and Services sales:

1. Retail sales:

It includes all durable and non durable goods sales

Retail Sales: A big component of GDP is Retail Sales.

This link describes Retail sales:

https://www.thebalance.com/what-is-retail-sales-3305722

So, you can think of retail sales as anything end consumers buy from a store as well as online including services as hair dressing, hotels, restaurants, bars, etc.

This link from USA census website provides details of retail sales starting from 1992.

https://www.census.gov/retail/marts/www/timeseries.html

This link shows Total Retail sales including everything (Stores selling physical goods (physical/online), gasoline, restaurants, Auto).

https://www.census.gov/retail/marts/www/adv44x72.txt

2019 Retail Categories: For 2019, Total Retail sales were $6.2T. There are 13 retail categories. Auto turns out to be biggest component of Retail sales. Following are few important categories for 2019.

1. Auto: For 2019, Auto sales were about $1.1T. This includes Auto services also. About 17M passenger vehicles got sold in USA in 2019, with an avg price of $35K. So, that would imply 17M*$35K = $600B in new car sales. Then there are used car sales, and auto parts sales and auto services, that probably account for remaining other $0.5T ($400B in used car sales + $100B in repairs as per this link: https://carsurance.net/blog/automotive-industry-statistics/ )
2. Online or mail only business (non-store): These were about $0.8T in 2019, a 10 fold rise from $0.08T in 1992. Online sales will pretty soon exceed Auto sales.
3. Food services, drinking places: In 2019, this totaled $0.8T
4. Grocery stores: About $0.7T for 2019
5. Gas: For 2019, Gas sales were about $0.5T. Assuming gas prices of $2/gallon, that implies sale of 0.25T gallons of gas used. With 200M vehicles on road, and each vehicle averaging 12K miles/year and let's assume 20miles/gallon as car gas consumption, that implies 12K/20 = 0.6K gallons used every year (or about 2 gallons a day which seems reasonable). So, total gasoline per year is easily 200M*0.6 = 120B gallons of gasoline consumed every year. Of course, a lot of these vehicles burn a lot more than 2 gallon a day (looks more like 4 gallons a day), since we are off by a factor of 2.

 If you look at sales data for 1992, retail sales were about $2T. So, in about 27 years, retails sales more than tripled to $6.2T.  What's confounding is that retail sales go up by 4% - 5% every year, even though people's wages are going up by 2% a year (as per IRS tax returns). So, where do people get extra money to keep spending beyond their wage increase, year after year? Maybe it's the extra debt they take to keep spending more than what they earn. But then, the interest will start eating into their income, where they can't afford to take on any more debt. Not sure what's going on. We need to find that?

2. Services Sales:

Bea website shows total sales as $14.8T. "goods" consumption at $4.6T, but Auto sales show as $0.5T (not sure why, since retail sales show auto consumption at $1.1T). Then services consumption is at $10.2T. Part of these services (about $1.6T) was included in retail sales number above. So, we are left with about $8.6T in services that's not part of retail sales.

The main components of this services consumption of $8.6T is as below:

1. Housing rental cost: This, combined with utility bills, is the biggest component of services. For 2019, total rental expenditure was $2.3T. Rental was $0.6T, while rental equivalent of owner occupied housing was $1.7T. Since 40M households are renting, and assuming avg rental of $15K/year => Total rental = 40M*$15K = $0.6T. Also with 70M homeowners, and assuming $2K/month rental equivalent, we get about 70M*$2K*12=$1.7T. NOTE: this is rental equivalent, the actual cost of home owning is higher since property taxes, interest, insurance, hoa dues can easily amount to $3K/month (with avg home price of $400K, and interest of 3%, property tax of 1%, HOA+insurance of 1% and 3% principal payment, it's easily 8% of home price). NOTE: it's hard to find home owner's cost of owning house, since most of the people have houses that appreciated in price, rather than them buying houses at these high prices. In order to find total expenses for housing, we have to add all housing interest income for all US banks, then add total property tax collection for all houses in USA, then add home insurance income for all insurance companies in USA. We can't include principal payments of mortgages for used houses, since they show up as expense for buyer, but show up as -ve expense for seller (i.e income for seller), so it's just trading of houses between buyer and seller. We have to include revenue from sale of new houses, since these are are the ones that got added to economy. We have to include price appreciation of used houses though. FIXME? Not sure how to include these? FIXME
2. Housing Utilities: These expenses are around $0.35T with Electricity=$0.2T, Water=$0.1T and Gas=$0.05T
3. HealthCare: These amounted to $2.5T for doctors visits, hospital bills, nursing home, etc. This is the 2nd biggest component of services. This doesn't include your health premium, as that's included in separate "insurance" category below.
4. Financial services: These amounted to $0.75T. These include bank fees, brokerage commissions, trading fees, mutual fund charges, etc
5. Insurance services: These were about $0.45T. These include health insurance, life insurance, auto insurance and home insurance.
6. Personal Communication services: These add to about $0.35T.  These include Cable/Satellite services ($0.1T), Cell phone ($0.14T), Internet ($75B), landline($25B)
7. Education services: Cost of college, schools etc amounts to $0.3T
8. Professional services: These add to about $0.2T. These include legal services ($0.1T), accounting services (tax, etc = $0.05T), various organization dues, burial services, etc.
9. Public transportation: These amount to $0.2T. These include air ($0.12T), road (cab, bus, etc amounting to $0.06T), rail and water transportation.
10. Misc expenses: Remaining $1.2T is in many misc categories as social services, religious activities, domestic household services, etc. These also include expenses of non profit institutions which is about $0.5T. Non profit org have gross output of $1.7T and receipt collections of $1.2T, resulting in final consumption expenditure of $0.5T. The reason I think we include expenses of non profit orgs as consumer expenses is not sure. ?FIXME??

Details

Published: Sunday, 20 September 2020 15:46

Hits: 1854

Put details

Details

Published: Thursday, 17 September 2020 19:26

Hits: 1961

scikit-learn:

It's an open source machine learning library for python. It's built on on top of SciPy and is distributed under the 3-Clause BSD license. The project was started in 2007 by David Cournapeau as a Google Summer of Code project, and since then many volunteers have contributed. scikit-learn is also known as sk-learn and provides simple and efficient tools for data mining and data analysis. It supports supervised and unsupervised learning. It also provides various tools for model fitting, data preprocessing, model selection and evaluation, and many other utilities.

Offical website is:

https://scikit-learn.org/stable/

Install on CentOS 7:

scikit-learn requires:

• Python (>= 3.6)
• NumPy (>= 1.13.3)
• SciPy (>= 0.19.1)
• joblib (>= 0.11)
• threadpoolctl (>= 2.0.0)

Scikit-learn plotting capabilities (i.e., functions start with plot_ and classes end with “Display”) require Matplotlib (>= 2.1.1). So, before you install scikit-learn, you need to have Numpy, SciPy and Matplotlib installed. scikit-learn may install it for you if it finds them missing. It does install other nodules for you as well.

Scikit-learn 0.20 was the last version to support Python 2.7 and Python 3.4. scikit-learn 0.23 and later require Python 3.6 or newer. As we will work with python3.6, we'll install scikit-learn 0.23-2 which is the latest version.

Cmd: run below cmd on Linux Terminal:

sudo python3.6 -m pip install -U scikit-learn

Screen messages:

We see following on screen: It first downloads scikit-learn-0.23-2, then it looks for scipy version >= 0.13.3, numpy version >= 1.8.2, and few other python modules. It downloads ones that are needed. It uninstalls ones that are older and replaces them with newer version. As as ex, below I had numpy-1.19.1 installed, but scikit-learn had latest numpy-1.19.2 version, so it uninstalled the older version, and replaced it with newer version.

  Downloading https://files.pythonhosted.org/packages/5c/a1/273def87037a7fb010512bbc5901c31cfddfca8080bc63b42b26e3cc55b3/scikit_learn-0.23.2-cp36-cp36m-manylinux1_x86_64.whl (6.8MB)

Collecting numpy>=1.13.3 (from scikit-learn)
  Downloading https://files.pythonhosted.org/packages/b8/e5/a64ef44a85397ba3c377f6be9c02f3cb3e18023f8c89850dd319e7945521/numpy-1.19.2-cp36-cp36m-manylinux1_x86_64.whl (13.4MB)

Collecting scipy>=0.13.3 (from scikit-learn)
  Using cached https://files.pythonhosted.org/packages/14/92/56dbfe01a2fc795ec92b623cb39654a10b1e9053db594f4ceed6fd6d4930/scipy-1.2.3-cp34-cp34m-manylinux1_x86_64.

Requirement already up-to-date: scipy>=0.19.1 in /usr/local/lib64/python3.6/site-packages (from scikit-learn)
Installing collected packages: joblib, numpy, threadpoolctl, scikit-learn
  Found existing installation: numpy 1.19.1
    Uninstalling numpy-1.19.1:
      Successfully uninstalled numpy-1.19.1
Successfully installed joblib-0.16.0 numpy-1.19.2 scikit-learn-0.23.2 threadpoolctl-2.1.0

Once we see above sucess message, That means scikit-learn is installed on your system. As explained in"modules" section, if the module gets installed correctly, we will see the module in below dir for python3.6:

/usr/local/lib64/python3.6/site-packages/sklearn => This is the scikit-learn dir. We also see a scikit-learn.libs dir which has *.so file (shared object library) and a scikit_learn-0.23.2.dist-info dir, which has all distribution info.

In order to check your installation and to see which version and where scikit-learn is installed, use below cmd:

> python3.6 -m pip show scikit-learn => It gives below o/p showing scikit-learn version 0.23.2 is installed

Name: scikit-learn
Version: 0.23.2
Summary: A set of python modules for machine learning and data mining
Home-page: http://scikit-learn.org
Author: None
Author-email: None
License: new BSD
Location: /usr/local/lib64/python3.6/site-packages
Requires: joblib, threadpoolctl, numpy, scipy

Error Messages:

As explained in "modules" section, if we just type "pip install U scikit-learn", we'll get multiple errors (files not found, etc) as we are not running right version of pip for python 3.6. You may get any of these errors as shown below: (note that even though python3 is soft linked to python3.6, below cmds keep using python3.4. So, it's best to run pip with python3.6 as explained above, and you will get smooth installation)

numpy errors running with python3.4

Building wheels for collected packages: numpy
  Running setup.py bdist_wheel for numpy ... error
  Complete output from command /usr/bin/python3.4 ....

multiple gcc compile errors

  gcc -pthread _configtest.o -o _configtest
  _configtest.o: In function `main':
  /tmp/pip-install-r7v7kemj/numpy/_configtest.c:6: undefined reference to `exp'
  collect2: error: ld returned 1 exit status

  gcc: _configtest.c
  _configtest.c:1:20: fatal error: Python.h: No such file or directory
   #include <Python.h>

Usage:

import sklearn: We need to first import sklearn and other modules in any python pgm. These are the imported modules:

import numpy as np
import matplotlib.pyplot as plt
import sklearn

linear model: sklearn has built in regression models to find best fit for given data. More details here:

https://scikit-learn.org/stable/modules/linear_model.html

Linear Regression:

Here (X,Y) data is fitted using weight coefficients. Here Y may be single target, or Y may be multiple targets (i.e Y₀, Y₁, etc that we are trying to fit simultaneously). Usually Y is a single target for our purposes. Linear regression fits in a linear model to minimize sum of squares of error. LinearRegression will take in its fit method arrays X, y and will store the coefficients of the linear model in it's coef_ member and the bias (or intercept) in it's ntercept_ member. When y is a single target, _coeff is 1D ndarray of shape(num_of_features,), while _intercept is just a float number. When y is multiple target, then _coeff is 2D ndarray of shape(num_of_targets, num_of_features), while _intercept is 1D array of shape(num_targets,). fit(X,y) method takes in 2 arrays, where X is 2D array of shape(num of samples, multiple X attr as X₀, X₁, and so on). y is a 1D array of output values.

Ex: This tries to fit data (X,y) using Linear regression. y=m₀X₀ + m₁X₁ + b

from sklearn import linear_model
reg = linear_model.LinearRegression() #reg is an instance of LinearRegression class (See Object Oriented pgm)
reg.fit([[0, 0], [1, 1], [2, 2]], [0, 1, 2]) #Here X has 2 attr X0, X1, and for each [X0, X1] we have y. So, for X=[0,0], Y=0. Similarly for X=[1,1], Y=1 and so on.
print(reg.coef_, reg.intercept_) => returns array([0.5, 0.5]), 0.1*e-16 . These are the 2 coeff m0 and m1,and intercept b that try to fit the data. so, y=0.5*X0 + 0.5*X1 + b for best fit. b is close to 0 (ideally it should be 0, but computers can't get exact 0). Here _coeff is 1D array while intercept_ is a float as expected

Logistic Regression:

This is implemented in LogisticRegression() class. This implementation can fit binary, One-vs-Rest, or multinomial logistic regression with optional , or Elastic-Net regularization. The solvers implemented in the class LogisticRegression are “liblinear”, “newton-cg”, “lbfgs”, “sag” and “saga”.

LogisticRegressionCV implements Logistic Regression with built-in cross-validation support, to find the optimal C and l1_ratio parameters according to the scoring attribute.

ex: It fits data (X,Y) using Logistic Regression where Y=0 or 1 for any given X. X is 2D array, while Y is 1D array, same as in previous linear regression example. The difference is that _coeff and _intercept now are diff shape matrix. _coeff is 2D ndarray of shape(1, num_of_features), while _intercept is 1D array of shape(1,).  Not sure why the matrix are higher dimensions now, even though the data for m, b that they contain is still same style as linear regression.

clf = sklearn.linear_model.LogisticRegressionCV();
clf.fit(X, Y);

print(clf.coef_, clf.intercept_) => prints coefficient matrix + bias (intercept) for the mode that fits this data closest. Prints something like: coeff=[[ 0.02783873 -0.20163637]] intercept=[0.01543046]. NOTE: ceff is 2D array, while intercept is 1D array (different than Linear Regression)

LR_predictions = clf.predict(X) => We can use predict method and apply it on original X dataset to see what predicted Y array it gives out. coefficients stored in clf.coef_ are used for predict method.

LR_predict_probability = clf.predict_proba(X) => This shows the probability for each example in X dataset. It shows it as a pair, where 1st num is probability of matching, while 2nd num is probability of not matching