Nevin Manimala

AMAZON IS USED to fielding accusations: that it has killed off physical retail business, that it mistreats warehouse workers, that it abuses its dominant platform in online sales. So perhaps it is not a surprise that some people also blame it for low inflation. In 2017 Janet Yellen, then chair of the Federal Reserve, wondered aloud if cut-throat online competition might be stopping goods-producers raising prices even in a world of rising demand. Alberto Cavallo of Harvard Business School has found that Amazon’s prices are 6% lower than those of eight large retailers, and 5% lower than on those retailers’ websites. The internet in general is no place to go in search of inflation: in America online prices have been falling fairly steadily since about 2012 and are lower than they were at the turn of the millennium.

Yet the so-called “Amazon effect” should not seem so novel. The winds of disinflation have been blowing through American retail for decades. In the 1990s and 2000s big-box retailers like Walmart and Target ruthlessly cut goods prices as they optimised their supply chains. Cheap imports from China and other emerging-market economies squeezed domestic producers. One study in 2008 found that low-wage countries capturing 1% of market share in America was associated with a 3.1% fall in producer prices. There has been barely any cumulative rise in American consumer-goods prices, excluding food and energy, for two decades. Before the financial crisis, inflation as a whole behaved normally because services inflation held up. Today, both goods and services inflation are low (see chart). The rise of online retail does not easily explain that broader shift.

Nonetheless, technological advance is a disinflationary force worth pondering. At a basic level, it allows an economy to produce more with its finite resources. If aggregate demand does not keep up, prices will fall—or at least not rise as fast. The idea that inflation has been low lately because productivity growth has been strong seems laughable everywhere except Silicon Valley because economic statistics have documented a global slowdown in productivity growth. Yet there is an argument that statisticians fail to capture some technological advances, making productivity seem lower and inflation higher than they really are.

The basic concern is a longstanding one. Because it takes a while for statisticians to notice that consumers are buying new products, they miss precipitous price falls early in a product’s life. It is also hard to tell how much better new products are than what went before. In today’s economy the missed value comes from smartphones, social media and online streaming. Spencer Hill, an economist at Goldman Sachs, recently calculated that the measured growth in consumption of personal electronics, communications and media was lower in the 2010s than in any of the five preceding decades. That was despite the fact that in 1990 it would have taken perhaps $3,000 to replicate even the basic functions of a modern phone—and only by using very bulky devices. In real terms, consumption in this category is surely soaring. The statistics must be missing something.

Statisticians are constantly battling the problem. But a review of America’s inflation indices in 2018 by Brent Moulson, a former top government official, estimated that the inflation index targeted by the Fed remained upwardly biased by almost half a percentage point, primarily because of new products and quality changes. The shift to online sales could be making new-product bias worse. A paper by Austan Goolsbee and Peter Klenow of Stanford University found that even excluding clothing, for which tastes are fickle, 44% of online sales in a database produced by Adobe Analytics, a computing company, were of goods that did not exist in the prior year. With such high churn the basket of goods monitored by official statisticians would quickly go stale. Messrs Goolsbee and Klenow have, for some categories of goods, helped Adobe Analytics to construct its own “digital price index” which shows much less inflation than official measures. For example, they find that furniture and bedding fell in price by almost 12% online between January 2014 and June 2019, while the official consumer price index records a fall of only 2.1%.

A bigger problem than falling prices is prices that are zero from the start. Most consumers today carry devices in their pockets with which they can make a video-call anywhere in the world, access information on any subject and translate languages instantaneously, all for free. The explosion in the provision of free services is usually cited as a reason to doubt the accuracy of GDP. But it is as big a problem for inflation. First, free services sometimes replace ones that were previously paid for, which puts new-product bias on steroids. Second, if consumers derive a greater share of their well-being from things that come free, inflation ceases to be a good measure of the cost of living or of the purchasing power of incomes.

The value of nothing

Measuring the price of something and measuring its value to consumers are two different tasks. Erik Brynjolffson of MIT and two co-authors have run experiments in an attempt to do the latter. They asked 3,000 online participants what they would need to be paid to give up Facebook for a month, offering to enforce the deal for a few randomly selected participants using Facebook features that reveal to friends when somebody last logged on. The median response was $42. About a fifth of users quoted somewhere near $1,000. In another experiment they struck similar agreements with participants at a Dutch university, enforcing the contract by getting users to change their passwords, in effect locking them out of their accounts, or to submit to monitoring of their electronic devices. The median figure participants quoted to give up mapping services for a month was about €59 ($64); for WhatsApp it was €536. In another paper Mr Brynjolffson and his colleagues asked consumers what they would need to be paid to forgo free online search engines for a year: the median response was over $17,500.

These figures can mislead. People will always fear the social isolation that would come with being cut off from the predominant communications technology of the day, whether it is telephones, texts or TikTok. Inflation and GDP were never intended to measure consumer welfare. Some free services are displacing activity which has never been counted in GDP, like casual matchmaking. Free services funded by advertising are not new: radio and television have been around a long time. And advertising is only small relative to the economy. John Fernald of the San Francisco Fed argues that many of the consumer benefits from modern technology are “conceptually non-market”.

Yet the line between market and non-market services is hazy. Imputed rent, the money homeowners would have to pay to rent a house equivalent to the one they own, is included in inflation and GDP, despite not representing any market transaction. In another recent paper David Byrne of the Federal Reserve and Carol Corrado of the Conference Board, a business group, argue that smartphones, broadband connections and Netflix subscriptions should be viewed as investments that reap variable dividends over time depending on how intensively they are used. Armed with trends in data usage and time-use surveys Mr Byrne and Ms Corrado construct a quality-adjusted price index for digital access services that shows prices falling by 21% between 2007 and 2017. The official price index for internet access, by contrast, shows prices up 4.5% over the same period.

The fact that inflation may be even lower than is reported is, in one respect, good news: it means that growth in living standards has been understated. But it is troublesome for central bankers who are already undershooting their inflation targets. Moreover, the justification for targeting inflation in the first place rests on the notion that the number is a meaningful representation of the economic experiences of the public and of firms. The more economic activity shifts into a domain where price is a slippery concept, the weaker that link will become. And there is another source of breakdown in economists’ understanding of how prices are formed: globalisation. ■

See previous article: Economists’ models of inflation are letting them down
See next article: Low inflation is a global phenomenon with global causes

Every 40 seconds, someone dies from suicide, and close to 800,000 fatalities occur each year in the United States. World Mental Health Day was established in 1992, with each year focusing on a different theme. This year’s focus is on mental health promotion and suicide prevention.

The World Health Organization (WHO) reports that 450 million people suffer from mental disorders. According to the National Alliance on Mental Illness (NAMI), one in five U.S. adults experiences mental illness, one in 25 experiences serious mental illness, and one in six U.S. youths ages 6 through 17 experiences mental illness.

In addition, over 70 percent of the youths in the juvenile justice system have been diagnosed with mental health issues, and over 41 percent of Veterans Health Administration patients have been diagnosed with mental illness or substance abuse.

Thus, suicide and mental health continue to an urgent issue because they affect the young, the elderly, men, women and veterans, among others. There has been a 30 percent increase in suicide since 1999, NAMI reports.

Some physicians and counselors have found that people affected by mental illness may not be eager to tell anyone or to seek help because of denial and shame. However, there are recognizable signs and changes in behavior that indicate someone is suffering from such illness.

Suicide is the 10th leading cause of death in the United States, and studies show that almost half of adults (46.4 percent) will experience a mental illness in their lifetime. Sadly, only 41 percent of people who had a mental disorder in the past year received professional health care or other services, according to NAMI.

In 2015, suicide and self-injury in the U.S. cost $69 billion, according to the American Foundation for Suicide Prevention. In 2017, 47,173 Americans died by suicide, and that year an estimated 1,400,000 suicide attempts occurred. The drive for mental health wellness, as well as advocacy and suicide prevention, will continue to be on the minds of physicians, counselors and the everyday people who are affected.

The WHO says that “a lack of urgency, misinformation, and competing demands are blinding policy-makers from taking stock of a situation where mental disorders figure among the leading cause of disease and disability in the world.”

Many hope that mental health awareness, and having mental health discussions, will help get those affected the help they need to live productive lives.

Meanwhile, resources are available for information on how to help someone with a mental illness. And remember, mental illness and suicide affect so many Americans that this is not just a day of observance but an effort to address a serious public health issue.

If you have thoughts of suicide, confidential help is available for free at the National Suicide Prevention Lifeline. Call 1-800-273-8255. The line is available 24 hours every day.

Purple ribbons promote domestic violence awareness

By Eileen Persike
Editor

Driving through downtown Rhinelander, purple ribbons – 35 of them – call out to passersby from the lampposts lining Brown Street. The ribbons are shouting to everyone who sees them to Remember My Name. Each ribbon bears the name of one of Wisconsin’s 35 homicide victims from domestic violence in 2018.

“The 35 victims received threats of violence, threats to kill, experienced stalking, strangulation, obsessive jealousy and sexual assault,” said Melissa Kropidlowski, the domestic violence program coordinator for Tri-County Council on Domestic Violence and Sexual Assault. “If your partner says they’re going to kill you, believe them.”

This is National Domestic Violence Awareness Month, and every October Tri-County Council uses the designation to remind the community that domestic violence happens, and the victims are people.

Kropidlowski has worked as a victim advocate for about a decade and said on average, 400 people walk through the doors at Tri-County who are involved in some sort of domestic violence situation. She said she’s hearing that the violence is getting worse.

“More volatile, more dangerous, a lot more cruelty to animals, a lot of threats with weapons, especially firearms, knives – the kitchen is the most dangerous place during a domestic violence incident,” said Kropidlowski. “We know that whoever wants to get a hold of a gun is going to get a hold of a gun regardless of what the laws are.”

According an annual report on domestic violence homicides compiled by End Domestic Abuse Wisconsin, firearms were used in 65% of the homicides last year, despite 29% of the perpetrators being legally prohibited from owning a gun.

Locally, the agency has a legal advocate who follows every domestic violence case and every sexual assault case from the initial appearance through sentencing. Kropidlowski said she is pleased to see more domestic violence cases go through the court system.

“The victims are now saying, ‘I want this person to be finally held accountable for what they’ve been doing,’” she noted. “We know it’s probably not the first time that something has happened, it’s probably been a lot of times, but they’ve finally been pushed to the point of needing to call for help.”

In a perfect world, Kropidlowski admitted, there would be more accountability for the perpetrators.

“I think our court system does a great job and I understand that it’s set up to make agreements and make deals, but I think that when you see a deferred prosecution agreement it does sometimes send the wrong message to the victim,” she said, adding that it’s her job to advocate for the client. “It would be nice to see an ending where the person does even get some jail time.”

A community candlelight vigil will be held 7 p.m. Friday, Oct. 11 at the Oneida County Courthouse where the names of all 35 victims will be read aloud. Anyone interested in participating by reading a name can contact Kropidlowski at 715-362-6841.

The National Coalition Against Domestic Violence started the “Remember My Name” project together with Ms. Magazine in 1994 to create a national registry of names in effort to increase public awareness of domestic violence deaths. For more information on the domestic violence homicide report, visit www.endabusewi.org.

MILWAUKEE — The Milwaukee Police Department announced on Tuesday, Oct. 8 a publicly-accessible database with eight major crime categories: Homicide, rape, robbery, aggravated assault, burglary, theft-larceny, motor vehicle theft, and arson. It’s an effort to provide another level of transparency when it comes to violent crime.

“From a public health perspective, data is extremely important,” said Reggie Moore with the City of Milwaukee’s Office of Violence Prevention. “The more that we understand about where things are happening, the more we can try to get ahead of it.”

Users can filter crime data based on location (police district, aldermanic district, and neighborhood), time period, and type of crime.

CLICK HERE to view the MPD crime database

Officials said the crime statistics are updated daily. However, they are subject to be changed at a later date based upon further investigation.

The database doesn’t include specific addresses and names of people involved to protect the identity of the individuals.

Moore called it an important educational tool.

“I think this is something the community has asked for,” said Moore.

It allows people to see what types of crimes are happening in their neighborhood day-to-day, year-to-year.

“This actually gives people an opportunity to dig into those numbers themselves,” said Alderman Cavalier Johnson. “My hope is that the general public actually utilizes this tool and sees what’s going on as it relates to public safety in their own neighborhoods and blocks.”

Thank goodness for Michigan and Iowa last Saturday because without them, the Battle for NU might have earned the distinction as the B1G-est game of the week. There were 19 punts between the two teams (Huskers -10; Wildcats – 9). Do you remember when we went two or three seasons without forcing Ohio State into a single punt?

OK, I’ll stop dredging up even worse memories.

This was a defensive slog, the likes of which we never expected when Bill Moos hired a young, bright offensive guru to rebuild Husker football. But, if there is one thing that Northwestern excels at, it is dragging you down into the mud and making you wrestle the pig in their slop.

There were 612 yards of total offense between NU and NU and 739 net punting yards.

I’M TRYING TO MAKE SENSE OF THAT TOO!

The abbreviated offensive stats weren’t even the result of a turnover-fest either. Northwestern three one interception. There were three fumbles in the game, but the offense recovered each time. It wasn’t even a sack-fest. Each team only gave up one sack (altogether, those sacks lost three yards). There were some tackles for loss (Nebraska had eight and Northwestern had 10) but altogether those TFL’s drove offenses back 39 yards.

If it wasn’t for Wan’Dale Robinson, Isaac Armstrong would be Nebraska’s offensive MVP.

Color Coded Pile of Numbers

It is tempting to look at the color-coded pile and assume that Nebraska is just an average team all around. There are a couple of things the offense/defense does well, a lot they do okay some of the time and a couple things they clearly don’t do well at all.

Remember, the offense is a unit largely devoid of seniors. We’ve got freshman, sophomore and juniors all trying to find their footing. The defense on the other hand is a senior-laden group finally operating with some consistency in coaching.

Just your friendly reminder that next season at this time, we will be wailing and gnashing our teeth over “WTH happened to Chinander’s guys?” instead of wailing and gnashing our teeth over “WTH happened to Frost’s offense?”

You’re welcome.

When it comes to individual statistics, only Wan’Dale Robinson really stood out. He had 123 receiving yards, 44 rush yards (6.3 yards/carry), and 19 kick return yards for a total of 186 all-purpose yards.

Adrian Martinez was the leading passer (13 of 20 for for 145 yards – 49 of that coming on one play).

Noah Vedral was the second leading rusher for the Huskers (7 carries for 33 yards; 4.7/carry). Those 33 yards were enough to make him the second highest on the team for all-purpose yards too. It was not a good day for the Husker offense against a stout Wildcat D. Injuries surely didn’t help either.

Defensively, Collin Miller tied Mohamed Barry with eight tackles apiece. Ben Stille had a good day starting in place of Khalil Davis as he notched 5 tackles, a half sack. a tackle for loss and a quarterback hurry.

Minnesota Color-Coded Pile of Numbers

The UNDEFEATED Minnesota Golden Gophers await the Huskers in the Battle for the Broken Chair (not trademarked by either athletic department of course). Be sure to donate to the cause – all money raised from Nebraska’s fans goes to the Team Jack Foundation. Let’s be sure the athletic department can’t ignore the most awesome trophy we have as a Big Ten team forever!

As you can see below, Minnesota is a decent team who doesn’t make silly mistakes. They are 5-0 against a not-too-difficult schedule (their two conference foes have been Illinois and Purdue and non-conference schedule was South Dakota State, Fresno State and Georgia Southern).

So, the Gophers haven’t played anyone yet? Husker fans have no room to crow since the Gophers can only play what’s on their schedule and they’ve won them all (some nail biters, but still “W’s”).

Remember when Ohio State came to Lincoln and there was a fair bit of “OSU has played a pretty soft schedule so far”. People were still saying that when OSU met Michigan State a week later.

It is up to Nebraska to be sure people aren’t still saying “they ain’t played nobody” when Minnesota meets…./checks Minnesota’s schedule….ummmm….Penn State on November 9. (Yes, Minnesota got both Rutgers (Oct 19) and Maryland (Oct 26) as crossover games with the East. That is part of the reason most of us at CN picked them for a fairly high finish in the West.)

Have a great week Corn Nation!

Descriptive Statistical Analysis helps you to understand your data and is a very important part of Machine Learning. This is due to Machine Learning being all about making predictions. On the other hand, statistics is all about drawing conclusions from data, which is a necessary initial step. In this post you will learn about the most important descriptive statistical concepts. They will help you understand better what your data is trying to tell you, which will result in an overall better machine learning model and understanding.

Table of Contents:

• Introduction
• Normal Distribution
• Central Tendency (mean, mode, median)
• Measures of Variability (range, interquartile range)
• Variance and Standard Deviation
• Modality
• Skewness
• Kurtosis
• Summary

Introduction

Doing a descriptive statistical analysis of your dataset is absolutely crucial. A lot of people skip this part and therefore lose a lot of valuable insights about their data, which often leads to wrong conclusions. Take your time and carefully run descriptive statistics and make sure that the data meets the requirements to do further analysis.

But first of all, we should go over what statistics really is:

Statistics is a branch of mathematics that deals with collecting, interpreting, organization and interpretation of data.

Within statistics, there are two main categories:

1. Descriptive Statistics: In Descriptive Statistics your are describing, presenting, summarizing and organizing your data (population), either through numerical calculations or graphs or tables.

2. Inferential statistics: Inferential Statistics are produced by more complex mathematical calculations, and allow us to infer trends and make assumptions and predictions about a population based on a study of a sample taken from it.

Normal Distribution

The normal distribution is one of the most important concepts in statistics since nearly all statistical tests require normally distributed data. It basically describes how large samples of data look like when they are plotted. It is sometimes called the “bell curve“ or the “Gaussian curve“.

Inferential statistics and the calculation of probabilities require that a normal distribution is given. This basically means, that if your data is not normally distributed, you need to be very careful what statistical tests you apply to it since they could lead to wrong conclusions.

A normal Distribution is given if your data is symmetrical, bell-shaped, centered and unimodal.

In a perfect normal distribution, each side is an exact mirror of the other. It should look like the distribution on the picture below:

You can see on the picture that the distribution is bell-shaped, which simply means that it is not heavily peaked. Unimodal means that there is only one peak.

Central Tendency

In statistics we have to deal with the mean, mode and the median. These are also called the „Central Tendency“. These are just three different kinds of „averages” and certainly the most popular ones.

The mean is simply the average and considered the most reliable measure of central tendency for making assumptions about a population from a single sample. Central tendency determines the tendency for the values of your data to cluster around its mean, mode, or median. The mean is computed by the sum of all values, divided by the number of values.

The mode is the value or category that occurs most often within the data. Therefore a dataset has no mode, if no number is repeated or if no category is the same. It is possible that a dataset has more than one mode, but I will cover this in the „Modality“ section below. The mode is also the only measure of central tendency that can be used for categorical variables since you can’t compute for example the average for the variable „gender“. You simply report categorical variables as numbers and percentages.

The median is the “middle” value or midpoint in your data and is also called the „50th percentile“. Note that the median is much less affected by outliers and skewed data than the mean. I will explain this with an example: Imagine you have a dataset of housing prizes that range mostly from $100,000 to $300,000 but contains a few houses that are worth more than 3 million Dollars. These expensive houses will heavily effect then mean since it is the sum of all values, divided by the number of values. The median will not be heavily affected by these outliers since it is only the “middle” value of all data points. Therefore the median is a much more suited statistic, to report about your data.

In a normal distribution, these measures all fall at the same midline point. This means that the mean, mode and median are all equal.

Measures of Variability

The most popular variability measures are the range, interquartile range (IQR), variance, and standard deviation. These are used to measure the amount of spread or variability within your data.

The range describes the difference between the largest and the smallest points in your data.

The interquartile range (IQR) is a measure of statistical dispersion between upper (75th) and lower (25th) quartiles.

While the range measures where the beginning and end of your datapoint are, the interquartile range is a measure of where the majority of the values lie.

The difference between the standard deviation and the variance is often a little bit hard to grasp for beginners, but I will explain it thoroughly below.

Variance and Standard Deviation

The Standard Deviation and the Variance also measure, like the Range and IQR, how spread apart our data is (e.g the dispersion). Therefore they are both derived from the mean.

The variance is computed by finding the difference between every data point and the mean, squaring them, summing them up and then taking the average of those numbers.

The squares are used during the calculation because they weight outliers more heavily than points that are near to the mean. This prevents that differences above the mean neutralize those below the mean.

The problem with Variance is that because of the squaring, it is not in the same unit of measurement as the original data.

Let’s say you are dealing with a dataset that contains centimeter values. Your variance would be in squared centimeters and therefore not the best measurement. 

This is why the Standard Deviation is used more often because it is in the original unit. It is simply the square root of the variance and because of that, it is returned to the original unit of measurement.

Let’s look at an example that illustrates the difference between variance and standard deviation:

Imagine a data set that contains centimeter values between 1 and 15, which results in a mean of 8. Squaring the difference between each data point and the mean and averaging the squares renders a variance of 18.67 (squared centimeters), while the standard deviation is 4.3 centimeters.

When you have a low standard deviation, your data points tend to be close to the mean. A high standard deviation means that your data points are spread out over a wide range.

Standard deviation is best used when data is unimodal. In a normal distribution, approximately 34% of the data points are lying between the mean and one standard deviation above or below the mean. Since a normal distribution is symmetrical, 68% of the data points fall between one standard deviation above and one standard deviation below the mean. Approximately 95% fall between two standard deviations below the mean and two standard deviations above the mean. And approximately 99.7% fall between three standard deviations above and three standard deviations below the mean.

The picture below illustrates that perfectly.

With the so-called „Z-Score“, you can check how many standard deviations below (or above) the mean, a specific data point lies. With pandas you can just use the „std()“ function. To better understand the concept of a normal distribution, we will now discuss the concepts of modality, symmetry and peakedness.

Modality

The modality of a distribution is determined by the number of peaks it contains. Most distributions have only one peak but it is possible that you encounter distributions with two or more peaks.

The picture below shows visual examples of the three types of modality:

Unimodal means that the distribution has only one peak, which means it has only one frequently occurring score, clustered at the top. A bimodal distribution has two values that occur frequently (two peaks) and a multimodal has two or several frequently occurring values.

Skewness

Skewness is a measurement of the symmetry of a distribution.

Therefore it describes how much a distribution differs from a normal distribution, either to the left or to the right. The skewness value can be either positive, negative or zero. Note that a perfect normal distribution would have a skewness of zero because the mean equals the median.

Below you can see an illustration of the different types of skewness:

We speak of a positives skew if the data is piled up to the left, which leaves the tail pointing to the right.

A negative skew occurs if the data is piled up to the right, which leaves the tail pointing to the left. Note that positive skews are more frequent than negative ones.

A good measurement for the skewness of a distribution is Pearson’s skewness coefficient that provides a quick estimation of a distributions symmetry. To compute the skewness in pandas you can just use the „skew()“ function.

Kurtosis

Kurtosis measures whether your dataset is heavy-tailed or light-tailed compared to a normal distribution. Data sets with high kurtosis have heavy tails and more outliers and data sets with low kurtosis tend to have light tails and fewer outliers. Note that a histogram is an effective way to show both the skewness and kurtosis of a data set because you can easily spot if something is wrong with your data. A probability plot is also a great tool because a normal distribution would just follow the straight line.

You can see both for a positively skewed dataset in the image below:

A good way to mathematically measure the kurtosis of a distribution is fishers measurement of kurtosis.

Now we will discuss the three most common types of kurtosis.

A normal distribution is called mesokurtic and has kurtosis of or around zero. A platykurtic distribution has negative kurtosis and tails are very thin compared to the normal distribution. Leptokurtic distributions have kurtosis greater than 3 and the fat tails mean that the distribution produces more extreme values and that it has a relatively small standard deviation.

If you already recognized that a distribution is skewed, you don’t need to calculate it’s kurtosis, since the distribution is already not normal. In pandas you can view the kurtosis simply by calling the „kurtosis()“ function.

Summary

This post gave you a proper introduction to descriptive statistics. You learned what a Normal Distribution looks like and why it is important. Furthermore, you gained knowledge about the three different kinds of averages (mean, mode and median), also called the Central Tendency. Afterwards, you learned about the range, interquartile range, variance and standard deviation. Then we discussed the three types of modality and that you can describe how much a distribution differs from a normal distribution in terms of Skewness. Lastly, you learned about Leptokurtic, Mesokurtic and Platykurtic distributions.

---

Niklas Donges is an entrepreneur, technical writer and AI expert. He worked on an AI team of SAP for 1.5 years, after which he founded Markov Solutions. The Berlin-based company specializes in artificial intelligence, machine learning and deep learning, offering customized AI-powered software solutions and consulting programs to various companies.