The Analytic Garden

Spring has actually arrived in upstate NY and along with it a small  COVID-19 surge.   First, here is the 7-day moving average of positive cases for the five county surrounding area. You can see a slight uptick in cases and hospitalization in recent days. Albany county has a larger population and shows a more dramatic increase. We have to keep in mind that the data is probably skewed due to fewer tests being reported. People with milder symptoms don't show up at testing centers of doctor's offices. Home tests aren't reported. Hospitalization data is more solid. Lineages In a  previous post , I commented on the difficulty of getting a clear picture of the state of the COVID-19 pandemic from publicly available data. Sequence and meta-data from  GISAID  is a rich and valuable source of information about the progress of the pandemic. Unfortunately, sequence data tends to lag case data. In addition, as public testing declines, opportunities to get viral material from...

My wife asked me to write a little program for her. She wanted to download some COVID-19 data from  health.data.ny.gov and calculate a moving average of the sum of two of the columns in the file. The URL of the data is  https://health.data.ny.gov/api/views/jw46-jpb7/rows.csv?accessType=DOWNLOAD . The data file has 37 columns and 148,011+ rows. We only care about the first few. Here is the header and the first four rows and several columns As of Date Facility PFI Facility Name DOH Region Facility County Facility Network NY Forward Region Patients Currently Hospitalized Patients Admitted Due to COVID Patients Admitted Not Due to COVID Patients Newly Admitted Patients Positive After Admission 3/26/2020 1 ALBANY MEDICAL CENTER HOSPITAL CAPITAL DISTRICT REGIONAL OFFICE ALBANY ALBANY MEDICA...

All models are wrong, but some are useful ~George E. P. Box All models are stories.  Is it our story that the pandemic is over? There's story that cases are down and symptoms are mild, so back to "normal". Is that reality? Does commonly available data tell a useful story? Can we build a useful model? Since the start of the pandemic, I have been looking at the positive cases in our local five county NY Capital District area. As did the rest of the country, ee saw a big spike in cases due to the Omicron variant. The data is from the NY Times . You can get the code here . The data shows a slight uptick in positive cases in recent days. Do these numbers reflect the current state of positive cases in the area? Probably not. Given high vaccination rates, the availability of home tests, and the the fact that most cases have mild symptoms, it's unlikely that the data reflects reality.  The linked article contains what to me is the most convincing evidence - a stark difference...

 I hesitate to call this deep learning . Anything running on a PC isn't all that deep . We looked at fitting this data using a neural net  previously  This time we will add a new set of data, global mean CO2 levels. In addition, instead of simple linear layers, we will use a fully recurrent neural network (RNN) as an initial layer.  The data we will analyze is from this post . It represents Temperature Anomaly  estimates  for the period 1850 to 2021 from the   NASA Global Climate Change   site .   We have seen this data before: here , here , and here . In addition, we will add a second source of data - the global mean CO2 concentration from 1850 through 2021 from Our World in Data . There are missing values in the downloaded data. They are replaced with linear interpolation using na.approx from the R zoo package. CO2_conc <- atmospheric_data %>% filter (Entity == 'World' & Year >= 1850 & Year <= 2...