Time Series Analysis: Final Project Dataset Presentation

Project

• Analyzing the amount of quarterly investment in technology companies, in the New York Metropolitan area, for 20 business sectors from 1995 to 2018.

Project Goal

• Determine what are going to be the ‘best’ sectors to enter into, as a data analyst/scientist, in the near future, if remaining in the NYC area.
• “best” defined as having the largest amount of projected future investment; translates to more opportunity for employment and greater potential for career growth.

Data

• 20 different sectors in tech industry
• 1 observation per quarter, 96 observations per sector

options(warn=-1)
library(ggplot2)
library(TSA)
library(readxl)
library(stringi)
library(zoo)

path = '/Users/mareksalamon/Desktop/School/Hunter/Spring Semester 2019/Time Series Analysis (715)/Final Project'
setwd(path)
data.nyc <- read_xlsx('Data/total_tech_invs_nyc.xlsx', col_names = TRUE, range = 'A2:C98',
                      col_types = c('text','numeric','numeric'), na = '0', trim_ws = TRUE)
data.nyc$Quarter <- as.Date(as.yearqtr(data.nyc$Quarter, format = "Q%q %Y"))
data <- read.csv('Data/all_tech_invs_nyc.csv')
head(data)

##      Quarter Number.of.deals Amounts      Sector
## 1 1995-01-01               0       0 agriculture
## 2 1995-04-01               0       0 agriculture
## 3 1995-07-01               0       0 agriculture
## 4 1995-10-01               0       0 agriculture
## 5 1996-01-01               0       0 agriculture
## 6 1996-04-01               0       0 agriculture

For simplicity, we will look at the total amount of quarterly investment across all 20 sectors.

head(data.nyc)

## # A tibble: 6 x 3
##   Quarter    `Number of deals`   Amounts
##   <date>                 <dbl>     <dbl>
## 1 1995-01-01                33 128280000
## 2 1995-04-01                22 108300000
## 3 1995-07-01                20  74240000
## 4 1995-10-01                22  74260000
## 5 1996-01-01                27 163330000
## 6 1996-04-01                25 166550000

Data Plot: Total Investment in NYC Tech Industry (1995-2018)

• No obvious seasonal trends
• Upward trend after 2010

Let’s take a look at the top 5 largest investments and smallest investments between 1995 and 2018.

# Top 5 largest investments
data.nyc[with(data.nyc,order(-Amounts)),][1:5,]

## # A tibble: 5 x 3
##   Quarter    `Number of deals`    Amounts
##   <date>                 <dbl>      <dbl>
## 1 2018-07-01               163 5806640000
## 2 2017-07-01               220 4526000000
## 3 2000-01-01               239 3983720000
## 4 2000-07-01               205 3379210000
## 5 2017-10-01               217 2978320000

# Top 5 smallest investments
data.nyc[with(data.nyc,order(Amounts)),][1:5,]

## # A tibble: 5 x 3
##   Quarter    `Number of deals`   Amounts
##   <date>                 <dbl>     <dbl>
## 1 1995-07-01                20  74240000
## 2 1995-10-01                22  74260000
## 3 1995-04-01                22 108300000
## 4 1995-01-01                33 128280000
## 5 1996-07-01                33 129000000

Data Plot: First Order Lag Plot of Quarterly Investments

• Vague positive, linear relationship
• Autoregressive model rather than sinusoidal model for best fit
• Not random

Data Plot: Autocorrelation of Quarterly Investment Amounts

• Bands linearly decrease with the lag
• Strong autocorreltion

# Autocorrelation function of the data
acf(ts(data.nyc$Amounts),main = "Correlogram")